Coal mine safety analysis method and system based on big data analysis

Abstract

The invention relates to the technical field of coal mine safety evaluation, in particular to a coal mine safety analysis method and system based on big data analysis. The method comprises the following steps: acquiring coal mine engineering environment data and coal mine engineering geographic data, wherein the coal mine engineering environment data comprises high-frequency gas concentration data and ventilation state data; extracting geographic topological information of coal mine engineering geographic data, and integrating coal mine engineering environment data into a coal mine post-disaster gas field initial data set; performing heat and mass coupling diffusion analysis on the initial data set of the coal mine post-disaster gas field to generate a residual gas concentration inversion network; and carrying out residual gas spatial distribution inversion on the initial data set of the coal mine post-disaster gas field by utilizing the residual gas concentration inversion network to obtain a spatial concentration residual map of each gas component. According to the invention, through multi-source data fusion and heat and mass coupling inversion, high-precision dynamic monitoring of a coal mine post-disaster gas field and precise positioning of an explosion source are realized, and the real-time assessment capability of coal mine safety risks is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine safety assessment, and in particular to a coal mine safety analysis method and system based on big data analysis. Background Art

[0002] Early coal mine safety relied primarily on manual inspections and empirical judgment, resulting in limited monitoring coverage and poor early warning effectiveness. With the widespread adoption of sensor technology and automated equipment, coal mines have begun to incorporate real-time data acquisition systems, enabling online monitoring of key parameters such as gas concentration, mine pressure, and temperature, driving the digitalization of safety monitoring. In the era of big data, coal mine safety analysis has increasingly leveraged massive, multi-source, and multi-dimensional data, using data fusion, mining, and modeling techniques to enhance risk identification and early warning capabilities. Early applications of big data technology in coal mines focused on data storage and simple statistical analysis, making it difficult to deeply explore patterns in potential hazards. With the introduction of advanced algorithms such as machine learning and artificial intelligence, coal mine safety analysis has shifted from static monitoring to dynamic prediction and intelligent decision-making, enabling the precise location and trend prediction of potential accidents. However, traditional methods often rely on simplified models in gas diffusion modeling, making it difficult to accurately reflect thermal-mass coupling effects. This leads to large errors in the inversion of the spatial distribution of gas concentrations and difficulty in accurately locating the explosion starting point. These methods often rely on single indicators and are susceptible to noise, resulting in limited real-time assessment capabilities for coal mine safety risks. Summary of the Invention

[0003] Based on this, it is necessary to provide a coal mine safety analysis method and system based on big data analysis to solve at least one of the above technical problems.

[0004] To achieve the above objectives, a coal mine safety analysis method based on big data analysis is provided, the method comprising the following steps: Step S1: Acquire coal mine engineering environmental data and coal mine engineering geographic data, wherein the coal mine engineering environmental data includes high-frequency gas concentration data and ventilation status data; extract geographic topology information from the coal mine engineering geographic data, and integrate the coal mine engineering environmental data into an initial data set of coal mine post-disaster gas field; Step S2: performing a thermal-mass coupled diffusion analysis on the initial dataset of the gas field after the coal mine disaster to generate a residual gas concentration inversion network; using the residual gas concentration inversion network, performing a residual gas spatial distribution inversion on the initial dataset of the gas field after the coal mine disaster to obtain a spatial concentration residual map of each gas component; Step S3: extracting high-order concentration gradient boundary characteristic data of the spatial concentration residual map of each gas component, and performing reverse tracing calculation of the explosion starting point source on the high-order concentration gradient boundary characteristic data, thereby generating an explosion point estimated coordinate domain; based on the explosion point estimated coordinate domain, predicting the spatial distribution evolution of the toxic gas residue on the spatial concentration residual map, and generating a toxic gas spatial distribution evolution map; Step S4: Couple and fuse the residual gas concentration inversion network, explosion point estimation coordinate domain, and toxic gas spatial distribution evolution map to construct a dynamic evolution map of air quality after a coal mine disaster; use the preset regional concentration threshold to conduct a safety risk assessment on the dynamic evolution map of air quality after a coal mine disaster to perform coal mine safety analysis operations.

[0005] Through coupled thermal-mass diffusion analysis and a residual gas concentration inversion network, this method accurately restores the spatial distribution of multi-component gases in complex post-disaster coal mine environments, providing a reliable data foundation for subsequent safety analysis. Reverse tracing based on high-order concentration gradient boundary features effectively improves the spatial positioning accuracy of the explosion starting point, providing important clues for disaster investigation and accident retrospective analysis. Predicting the spatial evolution of toxic gas residues using the estimated coordinate domain of the explosion point allows for early understanding of the diffusion direction and concentration trends of toxic gas in high-risk areas, providing a scientific basis for emergency rescue deployment. By coupling the inversion network, explosion coordinates, and evolution maps, dynamic spatiotemporal fusion of multi-source data is achieved, forming a comprehensive, three-dimensional, and visual air quality evolution map. Intelligent assessment of the air quality map using regional concentration thresholds allows for rapid identification of high-risk areas, enabling automated safety monitoring and risk warning response in coal mines. The entire method chain is logically rigorous, data-driven, highly scalable, and engineering-adaptable, effectively improving the technical level and response efficiency of post-disaster emergency management and safety assessment in coal mines. Therefore, the present invention realizes high-precision dynamic monitoring of the gas field after coal mine disasters and precise positioning of explosion sources through multi-source data fusion and thermal-mass coupling inversion, thereby improving the real-time assessment capability of coal mine safety risks.

[0006] Preferably, step S1 includes the following steps: Step S11: Acquire coal mine engineering environmental data and coal mine engineering geographical data, wherein the coal mine engineering environmental data includes high-frequency gas concentration data and ventilation status data; Step S12: resampling the high-frequency time series of the original coal mine engineering environment data to generate high-frequency gas concentration time data; Step S13: Divide the coal mine ventilation status data into regional ventilation modes to generate coal mine local ventilation topology data; Step S14: reconstructing the coal mine engineering geographic data into three-dimensional topology to generate coal mine geographic topology grid data; Step S15: Perform spatiotemporal fusion interpolation on the high-frequency gas concentration time data and the coal mine local ventilation topology structure data to generate coal mine ventilation-gas joint field data; integrate the coal mine ventilation-gas joint field data and the coal mine geographic topology grid data into the initial data set of the coal mine post-disaster gas field.

[0007] The present invention can reconstruct and optimize the time dimension of the original gas concentration data through high-frequency time series resampling, enhance the time continuity and sampling density of the data, and provide more stable and complete input data for subsequent modeling. By dividing the regional ventilation mode through ventilation status data, the local ventilation topology of the coal mine is constructed, and the expression ability of the local dynamic changes of the ventilation system is improved, which is conducive to more realistic simulation of gas propagation paths and behaviors. Constructing three-dimensional geographic topology grid data helps to truly reflect the complexity of the underground spatial structure and provide a high-precision spatial framework support for gas diffusion analysis. The spatiotemporal fusion interpolation of high-frequency gas concentration time data and ventilation topology data is realized to form a more complete ventilation-gas joint field, which not only improves data consistency, but also enhances the dynamic understanding of gas in spatial and temporal dimensions. The three key elements of gas, ventilation and geographic information are integrated to form an initial data set of the coal mine post-disaster gas field, providing an integrated and high-quality data foundation for subsequent gas diffusion modeling, risk assessment and safety decision-making. The entire process is centered on data fusion and closely integrated with the actual environment of coal mine engineering, effectively improving the physical rationality and engineering practicality of subsequent model reasoning results, thereby enhancing the reliability and accuracy of post-disaster air quality assessment and accident tracing.

[0008] Preferably, step S2 includes the following steps: Step S21: using the heat-mass cooperative diffusion equation to estimate the heat-mass transfer coefficient of the initial data set of the gas field after the coal mine disaster; Step S22: performing dynamic diffusion field simulation on the initial data set of the coal mine post-disaster gas field through the heat-mass transfer coefficient to generate post-disaster gas residual diffusion evolution data; Step S23: performing residual fitting on the post-disaster gas residual diffusion evolution data to construct a residual gas concentration inversion network; Step S24: Use the residual gas concentration inversion network to perform spatial inversion calculation on the initial data set of the gas field after the coal mine disaster to generate three-dimensional reconstruction data of the residual gas concentration; perform multi-component spectrum separation and spatial cluster analysis on the three-dimensional reconstruction data of the residual gas concentration to generate a spatial concentration residual map of each gas component.

[0009] By using the heat-mass co-diffusion equation to calculate the heat-mass transfer coefficient, the present invention fully considers the impact of heat on gas diffusion behavior, making simulation results more consistent with the physical reality of complex post-disaster coal mine environments and effectively improving the accuracy and adaptability of the diffusion model. Dynamic diffusion field simulation based on the heat-mass transfer coefficient can capture the spatial and temporal evolution of post-disaster gases, generating predictive residual diffusion evolution data and providing a realistic evolutionary basis for subsequent inversion and prediction. By performing residual fitting on the residual diffusion evolution data, an inversion network for residual gas concentration is established, improving the model's inversion accuracy in complex nonlinear gas propagation environments and enabling the inference of gas sources and concentration distributions from residual distributions. Spatial inversion using the inversion network generates three-dimensional reconstructions of residual gas concentrations, constructing high-resolution spatial gas concentration distributions and providing clear, visual three-dimensional data support for post-disaster assessments and emergency response. Multi-component spectral separation and spatial cluster analysis can identify and analyze the distribution patterns, aggregation areas, and diffusion boundaries of different gas types, generating spatial concentration residual maps for each gas component and providing fine-grained information for toxicity analysis and explosion risk assessment. The entire step chain is centered on the combination of physical model-driven and data-driven approaches, enabling efficient and automatic three-dimensional gas distribution reconstruction and component identification, significantly enhancing the technical depth and application value of post-disaster environmental modeling in coal mines.

[0010] Preferably, performing multi-component spectrum separation and spatial cluster analysis on the three-dimensional reconstructed data of residual gas concentration in step S24 includes: Perform Fourier transform on the three-dimensional reconstructed data of residual gas concentration, and extract the spectrum response characteristics of CH4, CO, CO2, and H2S after the transformation to generate spectrum response characteristic data; Blind source separation is performed on the three-dimensional reconstruction data of residual gas concentration using spectrum response characteristic data to extract the response signals of independent gases such as CH4, CO, CO2, and H2S, and generate independent gas component concentration data sets. Perform three-dimensional spatial cluster analysis on independent gas component concentration data sets to generate gas component spatial clustering map data; Perform geometric fitting and boundary reconstruction on the gas component spatial clustering map data to generate spatial concentration isosurface data; The concentration data of independent gas components and the isosurface map data are integrated to generate the spatial concentration residual map of each gas component.

[0011] By performing a Fourier transform on the three-dimensional reconstructed residual gas concentration data and extracting the spectral response characteristics of CH4, CO, CO2, and H2S, this method can accurately distinguish the frequency domain characteristics of each gas, significantly improving the separability between different gases and laying a solid foundation for subsequent separation. Using the spectral response characteristics for blind source separation, it can automatically extract independent gas component concentration data without the need for prior label information, avoiding errors caused by human intervention and enhancing the model's versatility and intelligence. Three-dimensional spatial clustering analysis of independent gas components can identify high-concentration areas and diffusion trends, revealing component distribution patterns within complex terrain structures and providing spatial guidance for accurate assessment of hazardous areas. By geometrically fitting and reconstructing the clustering results, concentration isosurface maps with clear boundaries and physical morphology are generated, achieving a transition from discrete point cloud data to continuous spatial representation and providing a high-quality geometric foundation for visualization and simulation analysis. Integrating the independent gas concentration data with the isosurface maps creates a spatial concentration residual map that comprehensively displays the local high-concentration areas, boundary trends, and changing trends of each gas, forming a core foundation for analysis such as toxic diffusion prediction and explosion source identification. The entire process is based on frequency domain analysis, enhanced by spatial clustering and geometric reconstruction, ultimately achieving intelligent deconstruction and precise modeling of complex post-disaster gas fields, significantly improving the scientificity, efficiency, and practicality of post-disaster gas distribution analysis in coal mines.

[0012] Preferably, step S3 includes the following steps: Step S31: calculating the local concentration change rate of each type of harmful gas in the spatial concentration residual map, and extracting high-order concentration gradient boundary feature data of the spatial concentration residual map based on the local concentration change rate; Step S32: performing toxic gas distribution analysis on each type of harmful gas in the spatial concentration residual map based on the local concentration change rate to construct a toxic gas distribution map; reconstructing the anti-diffusion vector field of the high-order concentration gradient boundary feature data to generate multi-directional equipotential vector tracking data; Step S33: performing a shock wave-ventilation path collaborative propagation path simulation on the high-order concentration gradient boundary characteristic data based on the multi-directional equipotential vector tracking data, and confirming the location of the initial energy source of the explosion by combining the simulation results with the concentration peak sequence and diffusion reverse vector trajectory of the spatial concentration residual map to generate an estimated coordinate domain of the explosion point; Step S34: Couple the explosion point estimated coordinate domain with the toxic gas distribution map to perform diffusion modeling to generate a time-varying initial diffusion field of the toxic gas; perform time-stepping simulation based on the time-varying initial diffusion field data of the toxic gas to generate a spatial distribution evolution diagram of the toxic gas residue.

[0013] The present invention accurately captures the gas concentration mutation area by calculating the local concentration change rate of harmful gases and extracting the high-order concentration gradient boundary features, constructing a spatially sensitive area that is more consistent with the physical characteristics of gas diffusion after the explosion, and providing higher-precision initial information for tracing and positioning. The toxic gas distribution analysis is carried out based on the concentration change rate. The generated toxic gas distribution map can clearly present the high-risk area and its evolution trend, further support the subsequent multi-field coupling modeling and dynamic diffusion simulation, and improve the authenticity and foresight of the accident scene simulation. By reconstructing the reverse diffusion vector field of the high-order gradient features, multi-directional equipotential vector tracking data simulating the explosion shock wave and the toxic gas diffusion path is constructed, reflecting the actual diffusion direction and path when the accident occurs, and improving the physical consistency and tracing reliability of reverse reasoning. Based on the simulation of the shock wave-ventilation path collaborative propagation path, the influence of the coal mine ventilation layout and structure on gas diffusion is comprehensively considered, so that the explosion point tracing analysis not only stays at the static reverse path calculation, but also integrates the environmental dynamic characteristics, significantly improving the authenticity and constraint of the explosion point estimation coordinate domain. By coupling the estimated coordinate domain of the explosion point with the toxic gas distribution map, constructing the time-varying diffusion initial field of the toxic gas, and performing time-stepping simulation, the residual trajectory of the toxic gas in complex space can be dynamically evolved in real time, providing refined risk evolution prediction support for emergency response. The generated toxic gas residual spatial distribution evolution map not only has high spatial resolution, but also can reflect the gas diffusion status at different time points, providing a highly reliable decision-making basis for formulating post-disaster ventilation plans, personnel search and rescue routes, and dangerous area blockades.

[0014] Preferably, step S33 includes the following steps: Step S331: performing regional shock wave response sensitivity division on the high-order concentration gradient boundary feature data to generate shock wave action weight distribution data; Step S332: performing main vector ventilation guide identification on the multi-directional equipotential vector tracking data to generate ventilation dominant path characteristic data; performing joint path mapping on the shock wave action weight distribution data and the ventilation dominant path characteristic data to generate shock wave-ventilation coordinated propagation simulation data; Step S333: extracting the concentration peak sequence of the spatial concentration residual map to generate explosion feature point sequence data; performing diffusion reverse vector fitting on the explosion feature point sequence data to generate explosion energy reverse propagation trajectory data; Step S334: performing path intersection analysis on the shock wave-ventilation cooperative propagation simulation data and the explosion energy reverse propagation trajectory data to generate an explosion point estimation coordinate domain.

[0015] The present invention divides regional shock wave response sensitivities based on high-order concentration gradient boundary feature data, quantifies the degree of response of different regions to explosion shock waves, and generates shock wave action weight distribution data. This allows explosion simulation to not only consider geometric paths but also accurately reflect the non-uniform propagation characteristics of explosion wave energy in space. Main vector ventilation guidance identification is performed on the equipotential vector field, and the characteristics of the dominant ventilation path are extracted. This allows the propagation path of toxic gas or shock waves to fully consider the actual airflow distribution and ventilation organization structure, thereby enhancing the simulation capability of propagation paths under actual coal mine ventilation conditions. By coupling the shock wave action weight data with the dominant ventilation path feature data, shock wave-ventilation collaborative propagation simulation data is formed, achieving the first "dual-field collaborative" path modeling of post-disaster explosion diffusion, significantly improving simulation accuracy and physical rationality. By extracting the concentration peak sequence and generating the explosion feature point sequence data, which is then used for reverse diffusion vector fitting, the energy diffusion path of the explosion source can be effectively restored, forming explosion energy reverse propagation trajectory data with strong physical logic, further enhancing the scientific basis for tracing the source. Through steps, the path intersection analysis of the shock wave-ventilation co-propagation simulation data and the explosion energy back-propagation trajectory data is carried out, and the two independent but complementary propagation mechanisms are integrated to generate the explosion point estimation coordinate domain, which significantly improves the spatial resolution and positioning accuracy of the explosion source position judgment.

[0016] Preferably, performing main vector ventilation guide identification on multi-directional equipotential vector tracking data includes: Perform vector direction consistency cluster analysis on multi-directional equipotential vector tracking data to generate equipotential vector main direction cluster data; evaluate the wind flow coupling intensity of equipotential vector main direction cluster data to generate ventilation adaptability matching score data; Perform multi-scale path topology mapping on the ventilation adaptability matching score data to generate ventilation path fitting grid data; perform time-series wind speed dynamic playback analysis on the ventilation path fitting grid data to generate dynamic ventilation flow trend map data; The dynamic ventilation flow trend map data is analyzed by combining path stability and dominance weight to generate ventilation dominant path characteristic data.

[0017] The present invention can effectively extract airflow paths with similar directions in space by performing vector direction consistency cluster analysis on multi-directional equipotential vector tracking data, generate equipotential vector main direction cluster data, greatly reduce the impact of multipath disturbance on the judgment of the main propagation direction, and improve the accuracy and robustness of ventilation guide identification. On the basis of the main direction of the equipotential vector, the wind flow coupling intensity evaluation method is introduced to generate ventilation adaptability matching score data, quantify the degree of coupling between different direction vectors and the actual ventilation system, and solve the technical shortcomings of the traditional method that cannot measure the degree of fit between the gas migration direction and the ventilation system. By performing multi-scale path topology mapping on the ventilation adaptability score data, the fitting path structure under different ventilation levels can be restored, and ventilation path fitting grid data can be generated to realize digital modeling of complex ventilation networks in coal mines, with strong ventilation path adaptability. Based on the fitting grid, time-series wind speed dynamic playback analysis is further carried out to generate dynamic ventilation flow trend map data, which can not only simulate the instantaneous airflow distribution, but also restore the diffusion path change process of the gas at different times, which is convenient for post-disaster rescue decision-making and ventilation optimization simulation. By integrating path stability and dominant weight analysis, the key factors affecting the gas propagation path are extracted, and finally highly directional ventilation dominant path characteristic data are generated, providing high-quality path basis for subsequent post-disaster gas tracking and tracing.

[0018] Preferably, the time-stepping simulation based on the initial field data of the time-varying diffusion of the poisonous gas in step S34 includes: Based on the time-varying diffusion initial field data of poisonous gas, multi-parameter heat and mass transport coefficients are extracted to generate diffusion evolution control parameter set data. The multi-parameter heat and mass transport coefficients include the mass diffusion coefficient, thermal diffusion coefficient and concentration gradient change rate of each gas. Initialize the three-dimensional multi-physics field time series grid of the diffusion evolution control parameter set data to generate the time-stepping cell data of the toxic gas diffusion; perform dynamic time series solution based on the steady state and boundary perturbations on the toxic gas diffusion time-stepping cell data to generate the multi-time concentration evolution field data of the toxic gas; Perform high-order spatial interpolation and continuity enhancement processing on the multi-time concentration evolution field data of poison gas to generate a time-continuous evolution map of poison gas concentration; The regional residual intensity cluster analysis is performed on the time-continuous evolution map of the toxic gas concentration to generate the spatial distribution evolution map of the toxic gas residue.

[0019] The present invention extracts multi-parameter heat and mass transport coefficients such as mass diffusion coefficient, thermal diffusion coefficient and concentration gradient change rate from the initial field of time-varying diffusion of poisonous gas, establishes a complete set of diffusion evolution control parameter data, realizes the real physical field driven simulation of poisonous gas diffusion behavior, and breaks through the bottleneck of rough approximation of traditional single diffusion model. By using the diffusion evolution control parameter set data to initialize the three-dimensional multi-physics field time series grid, highly discretized poisonous gas diffusion time step cell data can be generated, supporting the subsequent joint dynamic evolution of multiple physical fields at each moment and each spatial unit, and improving the resolution and sensitivity of space-time coupling simulation. The steady state and boundary perturbation coupling mechanism is used for dynamic time series solution to simulate the non-steady-state response behavior of poisonous gas under boundary mutation or local perturbation, and generate multi-time concentration evolution field data of poisonous gas, which can be used to evaluate the fluctuation, peak migration and regional aggregation risks of poisonous gas during the diffusion process, and enhance the adaptability and early warning capabilities of post-disaster diffusion prediction. High-order spatial interpolation and continuity enhancement processing are performed on multi-time evolution data to generate a continuous time evolution map of toxic gas concentration with high temporal resolution and smoothness. This visually expresses the dynamic evolution trend of concentration gradients throughout the toxic gas diffusion process, enhancing the perceptibility of the data results and the effectiveness of decision support. Using the continuous evolution map to perform regional residual intensity cluster analysis can effectively identify areas of high concentration retention, difficult-to-diffuse and clear areas, or areas with multiple sources of gas overlap after long-term evolution. This generates a spatial distribution evolution map of toxic gas residues, providing precise support for the formulation of post-disaster clearance plans and the planning of personnel evacuation routes.

[0020] Preferably, step S4 includes the following steps: Step S41: performing spatial fitting reconstruction on the residual gas concentration inversion network to generate high-dimensional inversion matrix data of gas concentration; Step S42: superimposing the explosion point estimated coordinate domain and the gas concentration high-dimensional inversion matrix data on the local explosion disturbance response to generate disturbance-enhanced gas concentration field data; Step S43: performing a three-dimensional multi-time series field fusion on the disturbance-enhanced gas concentration field data and the toxic gas spatial distribution evolution diagram to generate a dynamic evolution diagram of air quality after the coal mine disaster; Step S44: Use the preset regional concentration threshold to perform regional concentration threshold judgment on the dynamic evolution map of air quality after the coal mine disaster, and generate dangerous section concentration exceeding limit identification data; perform spatial distribution clustering and time series evolution trend analysis on the dangerous section concentration exceeding limit identification data, and generate a coal mine safety risk level annotation map to perform coal mine safety analysis operations.

[0021] The present invention generates high-dimensional inversion matrix data of gas concentration by spatially fitting and reconstructing the residual gas concentration inversion network, which can break through the limitations of traditional low-dimensional approximate modeling, reconstruct a gas concentration field with rich spatial distribution details and higher resolution, improve the modeling ability of the microscopic distribution law of residual toxic gas, and lay a data foundation for subsequent dynamic analysis. The explosion point estimated coordinate domain is superimposed with the high-dimensional concentration matrix data for local disturbance response to construct disturbance-enhanced gas concentration field data, which can dynamically reflect the secondary excitation and aggregation effects of the initial disturbance of the explosion on the distribution of toxic gas, and improve the simulation accuracy and sensitivity of the abnormal trend of the contaminated area after the disaster. The disturbance-enhanced concentration field is fused with the toxic gas spatial evolution map in three-dimensional multi-time series to construct a dynamic evolution map of air quality after the coal mine disaster, realize the joint tracking and trend visualization of toxic gas diffusion on the time scale and spatial scale, and comprehensively improve the reliability of the dynamic assessment of environmental risks after the accident. Using preset regional concentration thresholds to identify air quality evolution patterns, the system generates data identifying concentration violations in hazardous areas. This allows for rapid identification of areas with excessive concentrations, persistent high-risk areas, and diffusion tail-end convergence points, enhancing the effectiveness and efficiency of post-disaster ventilation, containment management, and personnel evacuation. Spatial distribution cluster analysis and temporal evolution trend analysis are performed on the concentration violation identification data in hazardous areas to create a coal mine safety risk level annotation map. This system supports the establishment of an intelligent, dynamic risk assessment and grading mechanism based on comprehensive indicators such as spatial distribution, concentration evolution, and duration, assisting coal mine safety command and decision-making.

[0022] In this specification, a coal mine safety analysis system based on big data analysis is provided, which is used to execute the above-mentioned coal mine safety analysis method based on big data analysis. The coal mine safety analysis system based on big data analysis includes: The gas field identification module is used to obtain coal mine engineering environmental data and coal mine engineering geographic data, where the coal mine engineering environmental data includes high-frequency gas concentration data and ventilation status data; extract the geographic topology information of the coal mine engineering geographic data, and integrate the coal mine engineering environmental data into the initial data set of the coal mine post-disaster gas field; The concentration analysis module is used to perform thermal-mass coupled diffusion analysis on the initial data set of the gas field after the coal mine disaster and generate a residual gas concentration inversion network. The residual gas concentration inversion network is used to invert the spatial distribution of residual gas on the initial data set of the gas field after the coal mine disaster and obtain the spatial concentration residual map of each gas component. The distribution evolution module is used to extract high-order concentration gradient boundary characteristic data from the spatial concentration residual map of each gas component, and perform reverse tracing calculation of the explosion starting point source on the high-order concentration gradient boundary characteristic data to generate the explosion point estimated coordinate domain; based on the explosion point estimated coordinate domain, the spatial concentration residual map is used to predict the spatial distribution evolution of the toxic gas residue to generate a toxic gas spatial distribution evolution map; The safety assessment module is used to couple and fuse the residual gas concentration inversion network, the explosion point estimation coordinate domain, and the toxic gas spatial distribution evolution map to construct a dynamic evolution map of air quality after a coal mine disaster; and to use preset regional concentration thresholds to conduct a safety risk assessment on the dynamic evolution map of air quality after a coal mine disaster to perform coal mine safety analysis operations.

[0023] The beneficial effect of the present invention is that the gas field identification module realizes the systematic collection and integration of complex environmental data after coal mine disasters by integrating high-frequency gas concentration data, ventilation status data and coal mine engineering geographical topology information, ensuring the comprehensiveness, timeliness and spatial accuracy of the initial data set of the gas field after the disaster, and providing a solid data foundation for subsequent diffusion analysis. The concentration analysis module relies on the thermal-mass coupling diffusion model to construct a residual gas concentration inversion network, which can accurately capture the multi-physical field diffusion characteristics of coal mine gas after the disaster, and realize high-precision inversion and component separation of the spatial concentration residual map of each gas component, effectively revealing the spatial distribution law of gas diffusion after the disaster. The distribution evolution module uses high-order concentration gradient boundary feature data, combined with the reverse tracing algorithm to complete the precise positioning of the explosion starting point, significantly improving the spatial accuracy of the explosion source estimation; based on the prediction of the spatial distribution evolution of toxic gas residues in the explosion point estimation coordinate domain, dynamic tracking and evolution simulation of the toxic gas diffusion process after the disaster are realized. The safety assessment module multi-dimensionally couples and fuses the residual gas concentration inversion network, the explosion point estimation coordinate domain, and the toxic gas spatial distribution evolution map to form a dynamic three-dimensional evolution map of post-disaster air quality, realizing real-time monitoring and trend prediction of the spatiotemporal changes of air pollution. By setting concentration thresholds for the dynamic evolution map to conduct risk assessment, rapid identification of dangerous areas and safety level division are achieved, providing scientific and quantitative decision support for coal mine safety management and emergency response, and significantly improving the efficiency and accuracy of coal mine post-disaster safety analysis. Therefore, the present invention realizes high-precision dynamic monitoring of the gas field after coal mine disasters and precise positioning of explosion sources through multi-source data fusion and thermal-mass coupling inversion, thereby improving the real-time assessment capability of coal mine safety risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of the steps of a coal mine safety analysis method based on big data analysis; Figure 2 for Figure 1 Detailed implementation steps of step S2 in FIG. Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG. The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0025] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0026] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0027] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0028] To achieve this, please refer to Figures 1 to 3 A coal mine safety analysis method based on big data analysis comprises the following steps: Step S1: Acquire coal mine engineering environmental data and coal mine engineering geographic data, wherein the coal mine engineering environmental data includes high-frequency gas concentration data and ventilation status data; extract geographic topology information from the coal mine engineering geographic data, and integrate the coal mine engineering environmental data into an initial data set of coal mine post-disaster gas field; Step S2: performing a thermal-mass coupled diffusion analysis on the initial dataset of the gas field after the coal mine disaster to generate a residual gas concentration inversion network; using the residual gas concentration inversion network, performing a residual gas spatial distribution inversion on the initial dataset of the gas field after the coal mine disaster to obtain a spatial concentration residual map of each gas component; Step S3: extracting high-order concentration gradient boundary characteristic data of the spatial concentration residual map of each gas component, and performing reverse tracing calculation of the explosion starting point source on the high-order concentration gradient boundary characteristic data, thereby generating an explosion point estimated coordinate domain; based on the explosion point estimated coordinate domain, predicting the spatial distribution evolution of the toxic gas residue on the spatial concentration residual map, and generating a toxic gas spatial distribution evolution map; Step S4: Couple and fuse the residual gas concentration inversion network, explosion point estimation coordinate domain, and toxic gas spatial distribution evolution map to construct a dynamic evolution map of air quality after a coal mine disaster; use the preset regional concentration threshold to conduct a safety risk assessment on the dynamic evolution map of air quality after a coal mine disaster to perform coal mine safety analysis operations.

[0029] Through coupled thermal-mass diffusion analysis and a residual gas concentration inversion network, this method accurately restores the spatial distribution of multi-component gases in complex post-disaster coal mine environments, providing a reliable data foundation for subsequent safety analysis. Reverse tracing based on high-order concentration gradient boundary features effectively improves the spatial positioning accuracy of the explosion starting point, providing important clues for disaster investigation and accident retrospective analysis. Predicting the spatial evolution of toxic gas residues using the estimated coordinate domain of the explosion point allows for early understanding of the diffusion direction and concentration trends of toxic gas in high-risk areas, providing a scientific basis for emergency rescue deployment. By coupling the inversion network, explosion coordinates, and evolution maps, dynamic spatiotemporal fusion of multi-source data is achieved, forming a comprehensive, three-dimensional, and visual air quality evolution map. Intelligent assessment of the air quality map using regional concentration thresholds allows for rapid identification of high-risk areas, enabling automated safety monitoring and risk warning response in coal mines. The entire method chain is logically rigorous, data-driven, highly scalable, and engineering-adaptable, effectively improving the technical level and response efficiency of post-disaster emergency management and safety assessment in coal mines. Therefore, the present invention realizes high-precision dynamic monitoring of the gas field after coal mine disasters and precise positioning of explosion sources through multi-source data fusion and thermal-mass coupling inversion, thereby improving the real-time assessment capability of coal mine safety risks.

[0030] In the embodiment of the present invention, reference Figure 1 FIG. 1 is a flow chart of the steps of a coal mine safety analysis method based on big data analysis according to the present invention. In this example, the coal mine safety analysis method based on big data analysis includes the following steps: Step S1: Acquire coal mine engineering environmental data and coal mine engineering geographic data, wherein the coal mine engineering environmental data includes high-frequency gas concentration data and ventilation status data; extract geographic topology information from the coal mine engineering geographic data, and integrate the coal mine engineering environmental data into an initial data set of coal mine post-disaster gas field; Step S2: performing a thermal-mass coupled diffusion analysis on the initial dataset of the gas field after the coal mine disaster to generate a residual gas concentration inversion network; using the residual gas concentration inversion network, performing a residual gas spatial distribution inversion on the initial dataset of the gas field after the coal mine disaster to obtain a spatial concentration residual map of each gas component; Step S3: extracting high-order concentration gradient boundary characteristic data of the spatial concentration residual map of each gas component, and performing reverse tracing calculation of the explosion starting point source on the high-order concentration gradient boundary characteristic data, thereby generating an explosion point estimated coordinate domain; based on the explosion point estimated coordinate domain, predicting the spatial distribution evolution of the toxic gas residue on the spatial concentration residual map, and generating a toxic gas spatial distribution evolution map; Step S4: Couple and fuse the residual gas concentration inversion network, explosion point estimation coordinate domain, and toxic gas spatial distribution evolution map to construct a dynamic evolution map of air quality after a coal mine disaster; use the preset regional concentration threshold to conduct a safety risk assessment on the dynamic evolution map of air quality after a coal mine disaster to perform coal mine safety analysis operations.

[0031] In this embodiment of the present invention, coal mine engineering environmental data is collected, including high-frequency gas concentration data (e.g., methane (CH4), carbon monoxide (CO), carbon dioxide (CO2), and hydrogen sulfide (H2S) concentrations that fluctuate after an explosion, obtained using laser spectral sensors and infrared sensors); and ventilation status data (e.g., fan operating status, wind speed sensor data, airflow direction, and cross-sectional ventilation parameters). The three-dimensional structure of the mine tunnels, ventilation paths, node topology, and their spatial relationships are extracted from the coal mine engineering geographic data. A graph topology modeling method is used to construct a coal mine spatial grid topology model, forming a geometric foundation for coupling environmental data. The high-frequency gas concentration data is spatially registered with the geographic topology. A multi-source data fusion algorithm (e.g., based on Bayesian estimation or Kalman filtering) is used to integrate ventilation status and gas data to generate an initial dataset of the coal mine post-disaster gas field at a unified timestamp. A thermal-mass coupled diffusion model is constructed, linking gas temperature gradients and mass diffusion based on the coupled derivation of the Navier-Stokes equations and Fick's diffusion law. The initial dataset of the coal mine post-disaster gas field is input and, combined with the ventilation disturbance function, the residual gas distribution equation is iteratively solved. A deep learning model, such as the Multiscale Convolutional Residual Network (ResUNet), is constructed for concentration inversion in the spatial and temporal dimensions. Network training samples can be generated based on simulations of historical coal mine accident scenarios. The output is a predicted concentration tensor for each gas, which is then inverted to generate residual concentration distribution results. The inversion results are mapped back onto the three-dimensional topological model of the mine tunnel to generate a spatial distribution map of gas components. The output includes spatial concentration residual maps of various gases, including CH4, CO, and H2S, at different time frames. The spatial concentration residual maps are processed using multi-order gradient operators (such as Laplacian of Gaussian and high-order Sobol filters). The boundaries of areas with sudden gas concentration changes are extracted as boundary features of the suspected detonation source. Using these characteristic boundary points, a vector direction field is back-traced. Using a gas source streamline inversion algorithm combined with the Backward Euler Approximation method, the possible paths of the detonation source are calculated. Cluster analysis is performed on the intersection of all paths to generate the estimated coordinate domain of the explosion point. Based on the estimated detonation coordinate domain and the current gas concentration state, a time-recursive gas transport equation is used for prediction. A moving source Gaussian PlumeModel is introduced to predict the toxic gas diffusion trajectory. A three-dimensional dynamic graph of the spatial concentration evolution of the toxic gas over time, known as the toxic gas spatial distribution evolution map, is output. The residual gas concentration inversion network and the estimated explosion point coordinate domain are coupled with the toxic gas evolution map. A dynamic evolution map of post-coal mine disaster air quality is constructed using multi-dimensional fusion algorithms (such as tensor fusion or graph neural networks (GNNs). The map supports querying air quality risk areas by time slice and supports multi-gas layered visualization.Set regional concentration thresholds (such as CH4>2%, CO>30ppm, etc.); perform threshold comparison analysis item by item for any spatial cell and time slice in the map; areas above the threshold are marked as "high-risk areas", and the system automatically generates a safety risk level map and corresponding emergency response strategy recommendations.

[0032] Preferably, step S1 includes the following steps: Step S11: Acquire coal mine engineering environmental data and coal mine engineering geographical data, wherein the coal mine engineering environmental data includes high-frequency gas concentration data and ventilation status data; Step S12: resampling the high-frequency time series of the original coal mine engineering environment data to generate high-frequency gas concentration time data; Step S13: Divide the coal mine ventilation status data into regional ventilation modes to generate coal mine local ventilation topology data; Step S14: reconstructing the coal mine engineering geographic data into three-dimensional topology to generate coal mine geographic topology grid data; Step S15: Perform spatiotemporal fusion interpolation on the high-frequency gas concentration time data and the coal mine local ventilation topology structure data to generate coal mine ventilation-gas joint field data; integrate the coal mine ventilation-gas joint field data and the coal mine geographic topology grid data into the initial data set of the coal mine post-disaster gas field.

[0033] In this embodiment of the present invention, a coal mine sensor network system is utilized to acquire coal mine engineering environmental data. Specifically, this data includes: high-frequency gas concentration data, collected by underground distributed sensors at a sampling frequency of no less than 1 Hz, covering key gases such as methane (CH4), carbon monoxide (CO), carbon dioxide (CO2), and hydrogen sulfide (H2S); and ventilation status data, which collects wind speed, pressure, direction, and fan operating status at each ventilation node. This data is also connected to the Coal Mine Geographic Information System (MineGIS) to export coal mine engineering geographic data, including shaft and tunnel structure, goaf, ventilation zones, and control point coordinates. Raw sensor data, due to noise and inconsistencies with sampling intervals, requires a unified time base. The raw gas concentration data is time-calibrated using an interpolation-based time resampling algorithm. Missing data periods are filled using linear or spline interpolation. The resampling ensures a consistent time step (e.g., 1s or 0.5s). High-frequency gas concentration time data is generated, represented as a gas concentration time tensor G(t, x, y, z). Based on ventilation status data and shaft and roadway spatial information, ventilation operation characteristics in different areas are identified. Clustering algorithms (such as K-means or DBSCAN) are used to classify wind speed and direction data into patterns, forming regional ventilation pattern labels. Ventilation patterns are embedded in the three-dimensional roadway network to construct local ventilation topology data for the coal mine: represented as a graph structure V = (N, E), where nodes N are ventilation control points and edges E are airflow channels, with weighted attributes (such as wind speed and ventilation pattern number) attached. Spatial reconstruction of coal mine engineering geographic data is performed, using voxel modeling or tetrahedral meshing for three-dimensional topological reconstruction. Using 3D Delaunay triangulation, the coal mine geographic topology grid data is constructed, with a multi-level nested mesh structure. Each grid cell has the following attributes: grid coordinate range; roadway type (main roadway, auxiliary roadway, return air roadway, etc.); adjacency topological relationships; and boundary attributes. The high-frequency gas concentration time data obtained in step S12 and the local ventilation topology obtained in step S13 are spatially interpolated. High-dimensional Kriging interpolation or inverse distance weighting (IDW) is used to expand the gas data within the ventilation topology to form multi-point spatiotemporal coupled data. A combined coal mine ventilation and gas field data set is constructed, represented as a five-dimensional tensor F(t, x, y, z, c), where c represents the gas type channel dimension. The combined field data F is spatially bound to a three-dimensional geographic topological grid. Using grid attribute indexing and topological constraints, the corresponding gas-ventilation information is injected into each geographic grid cell. The final output is an initial dataset of the coal mine post-disaster gas field, represented as the concentration change values ​​and ventilation vectors for each spatial cell (voxel) in the three-dimensional structure. This dataset possesses time series properties and can be used for subsequent simulation and inversion analysis.

[0034] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: using the heat-mass cooperative diffusion equation to estimate the heat-mass transfer coefficient of the initial data set of the gas field after the coal mine disaster; Step S22: performing dynamic diffusion field simulation on the initial data set of the coal mine post-disaster gas field through the heat-mass transfer coefficient to generate post-disaster gas residual diffusion evolution data; Step S23: performing residual fitting on the post-disaster gas residual diffusion evolution data to construct a residual gas concentration inversion network; Step S24: Use the residual gas concentration inversion network to perform spatial inversion calculation on the initial data set of the gas field after the coal mine disaster to generate three-dimensional reconstruction data of the residual gas concentration; perform multi-component spectrum separation and spatial cluster analysis on the three-dimensional reconstruction data of the residual gas concentration to generate a spatial concentration residual map of each gas component.

[0035] In this embodiment of the present invention, the diffusion characteristics of the gas field after a coal mine disaster are quantified to determine the speed and path of gas propagation under the coupled influence of high temperature and ventilation. First, a simulation scenario is established based on the coal mine's three-dimensional spatial grid and the initial gas concentration distribution. Temperature and gas concentration change data are collected at different times and locations after the explosion. Coupled analysis is performed using simulation algorithms (such as finite difference or finite element methods) based on the ventilation conditions in different areas of the coal mine. By comparing historical data with simulation results, the diffusion capacity of gas in different areas, namely the thermal diffusivity and concentration diffusivity, is estimated. Finally, a set of data describing the heat and mass transfer characteristics is generated for subsequent simulation analysis. Simulate the diffusion trajectory and evolution of residual harmful gases within coal mine passages and cavities after an explosion, applying the estimated diffusion coefficient data to the initial gas field. Using 3D gas simulation software or a custom model, factor in post-disaster ventilation changes and the presence of local heat sources. Initiate dynamic simulations, continuously advancing multiple time steps to observe how different gas components propagate and dilute in space. Simulation results are output as a time series, generating high-resolution residual gas diffusion evolution data. This data includes the evolution of concentration values ​​for various gases over time and space, useful for analyzing trends in high-risk areas. Build an intelligent model capable of predicting gas distribution beyond a certain point in time based on initial data, improving the accuracy of spatial distribution estimates. The simulation results were compared with actual sensor data to extract simulation error information. Based on this error information, a multi-layer three-dimensional neural network was constructed, consisting of an input layer, a convolution processing layer, a residual correction layer, and an output layer. The network was iteratively trained using the initial coal mine gas field as the model input and the error as the training target. After training, the inversion network was able to intelligently infer a more accurate spatial distribution of gas concentrations based on the initial environment. The model also exhibited generalization capabilities, adapting to different ventilation structures and gas composition combinations. The spatial distribution characteristics of different gas components were identified from the predicted three-dimensional gas distribution results, and their residual hotspots were analyzed. The inversion network output was used to reconstruct a spatial gas concentration map within the coal mine. For multiple gas components, a feature extraction algorithm was used to separate them and clarify the distribution characteristics of each gas in different regions. Spatial clustering analysis was performed on the distribution of each gas to identify high-concentration residual areas and their boundaries. During the clustering analysis, voxels were classified into different levels, and the correlation between adjacent high-concentration areas was identified. Finally, a three-dimensional residual concentration map of each gas component was generated, which was used for subsequent explosion point location and toxic gas evolution analysis.

[0036] Preferably, performing multi-component spectrum separation and spatial cluster analysis on the three-dimensional reconstructed data of residual gas concentration in step S24 includes: Perform Fourier transform on the three-dimensional reconstructed data of residual gas concentration, and extract the spectrum response characteristics of CH4, CO, CO2, and H2S after the transformation to generate spectrum response characteristic data; Blind source separation is performed on the three-dimensional reconstruction data of residual gas concentration using spectrum response characteristic data to extract the response signals of independent gases such as CH4, CO, CO2, and H2S, and generate independent gas component concentration data sets. Perform three-dimensional spatial cluster analysis on independent gas component concentration data sets to generate gas component spatial clustering map data; Perform geometric fitting and boundary reconstruction on the gas component spatial clustering map data to generate spatial concentration isosurface data; The concentration data of independent gas components and the isosurface map data are integrated to generate the spatial concentration residual map of each gas component.

[0037] In an embodiment of the present invention, frequency domain transformation is performed on the three-dimensional reconstructed data of residual gas concentration, and a three-dimensional Fourier transform method is used to perform frequency domain mapping on the data of gas concentration variations at each location as they change over time. The main frequency response characteristics of the target gases (CH4, CO, CO2, H2S) in the frequency domain, such as the main peak position, spectral amplitude, and frequency domain energy distribution, are extracted. A multidimensional feature vector dataset containing the spectral response characteristics is constructed to provide a basis for subsequent signal separation. Ultimately, spectral response feature data is generated, which can reflect the variation patterns of different gases in the spatial frequency domain. Blind source separation algorithms such as independent component analysis (ICA) are used to process the spectral response feature data. Without relying on a priori gas templates, the independent concentration signal exhibited by each gas in space is automatically extracted. A three-dimensional concentration dataset of independent gas components, such as CH4, CO, CO2, and H2S, is generated. This dataset has clear component attribution and spatial distribution characteristics, significantly improving the accuracy of subsequent cluster analysis. Based on each independent gas component concentration dataset, a density-based clustering method (such as DBSCAN and OPTICS) is used to perform three-dimensional spatial clustering. The cluster analysis fully considers the concentration similarity and spatial connectivity between neighboring voxels. The clustering centers, high-concentration distribution areas, and boundary structures of each gas type within the mine space are identified. The output is a gas component spatial cluster map data, which contains the geometric range and component characteristics of each cluster region. Each cluster region in the spatial cluster map is geometrically reconstructed. The isovalue concentration boundary of each cluster region is extracted using a three-dimensional isosurface extraction algorithm (such as Marching Cubes). The extracted isosurface is smoothed and structurally fitted to enhance boundary recognizability. Finally, a spatial concentration isosurface map data is generated, which shows the high, medium, and low concentration distribution of each gas in space. The concentration data of independent gas components are aligned and fused with the corresponding spatial isosurface maps; the concentration levels of different components are classified and labeled according to the concentration value range and spatial position; a spatial concentration residual map in a unified format is constructed with three-dimensional display capabilities and attribute query functions; each type of gas in the map has an independent level, clear structural boundaries, and a clear spatial change trend, which is suitable for subsequent spatial reasoning and evolution simulation.

[0038] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: calculating the local concentration change rate of each type of harmful gas in the spatial concentration residual map, and extracting high-order concentration gradient boundary feature data of the spatial concentration residual map based on the local concentration change rate; Step S32: performing toxic gas distribution analysis on each type of harmful gas in the spatial concentration residual map based on the local concentration change rate to construct a toxic gas distribution map; reconstructing the anti-diffusion vector field of the high-order concentration gradient boundary feature data to generate multi-directional equipotential vector tracking data; Step S33: performing a shock wave-ventilation path collaborative propagation path simulation on the high-order concentration gradient boundary characteristic data based on the multi-directional equipotential vector tracking data, and confirming the location of the initial energy source of the explosion by combining the simulation results with the concentration peak sequence and diffusion reverse vector trajectory of the spatial concentration residual map to generate an estimated coordinate domain of the explosion point; Step S34: Couple the explosion point estimated coordinate domain with the toxic gas distribution map to perform diffusion modeling to generate a time-varying initial diffusion field of the toxic gas; perform time-stepping simulation based on the time-varying initial diffusion field data of the toxic gas to generate a spatial distribution evolution diagram of the toxic gas residue.

[0039] In an embodiment of the present invention, the three-dimensional data of each type of hazardous gas in the spatial concentration residual map is traversed to calculate the concentration change rate per unit volume in the local voxel neighborhood. The change rate is spatially differentiated to identify high-gradient areas with significant concentration changes. High-order concentration gradient boundary features, including boundary shape, continuity, mutation points, and change direction, are extracted. The high-order concentration gradient boundary feature data is output for subsequent backdiffusion and vector tracking analysis. Based on the toxicological hazard level of each type of gas and the local concentration change rate, multiple risk concentration classification thresholds are set. A toxic gas distribution map is constructed, dividing the spatial range by toxicity level and marking key high-risk areas. The map contains attribute information such as three-dimensional position, concentration level, and distribution range. Vector backdiffusion modeling is performed on the extracted high-order concentration gradient boundary features. The gas source tracing path is simulated to construct a multi-directional equipotential vector field. Multi-directional equipotential vector tracing data is generated as a reference for subsequent energy propagation simulations. This multi-directional equipotential vector tracing data is coupled with the coal mine ventilation topology. The backpropagation trend of the shock wave along the main ventilation path is simulated. Ventilation factors such as wind speed, wind direction, and drag coefficient are considered during the simulation process. A co-propagation path diagram of the shock wave and ventilation path is output. The concentration peak sequence and its diffusion back vector trajectory are extracted from the spatial concentration residual map. This is spatially overlaid with the co-propagation path diagram to identify the overlap between the concentration peak source and the vector convergence point. An estimated explosion point coordinate domain is generated and output as a probability distribution range of the three-dimensional spatial location. The estimated explosion point coordinate domain is spatially coupled with the toxic gas distribution map. Initial boundary conditions such as initial concentration distribution, wind conditions, and structural obstacles are set. A time-varying initial diffusion field of the toxic gas is constructed, including a diffusion starting point, direction, and velocity model. The initial diffusion field of the toxic gas is simulated at discrete time steps to calculate the spatial distribution at different time points; the effects of ventilation disturbances, adsorption, and gas density differences on the propagation path are considered; a three-dimensional residual concentration map of the toxic gas is output at each time step to form a continuous evolution map; and finally a spatial distribution evolution map of the toxic gas residue is generated for safety assessment and emergency decision support.

[0040] Preferably, step S33 includes the following steps: Step S331: performing regional shock wave response sensitivity division on the high-order concentration gradient boundary feature data to generate shock wave action weight distribution data; Step S332: performing main vector ventilation guide identification on the multi-directional equipotential vector tracking data to generate ventilation dominant path characteristic data; performing joint path mapping on the shock wave action weight distribution data and the ventilation dominant path characteristic data to generate shock wave-ventilation coordinated propagation simulation data; Step S333: extracting the concentration peak sequence of the spatial concentration residual map to generate explosion feature point sequence data; performing diffusion reverse vector fitting on the explosion feature point sequence data to generate explosion energy reverse propagation trajectory data; Step S334: performing path intersection analysis on the shock wave-ventilation cooperative propagation simulation data and the explosion energy reverse propagation trajectory data to generate an explosion point estimation coordinate domain.

[0041] In this embodiment of the present invention, high-order concentration gradient boundary feature data is mapped onto a three-dimensional spatial grid. Each spatial region (voxel) is evaluated for shock wave response based on factors such as concentration mutation intensity, boundary continuity, and spatial topological complexity. The region is then divided into high-sensitivity, medium-sensitivity, and low-sensitivity regions according to their response level. Each region is assigned a corresponding shock wave action weight, forming a spatial hierarchy. Shock wave action weight distribution data is then output for collaborative ventilation path modeling. Multidirectional equipotential vector tracking data is clustered by vector direction. Paths with stable flow directions and significant flow velocities are selected based on ventilation system design drawings, ventilation velocity distributions, or sensor data. The dominant airflow paths in the ventilation system are extracted, and ventilation dominant path feature data is output. The shock wave action weight distribution data is spatially mapped and fused with the ventilation dominant path feature data. The fusion process considers the guiding effect of the dominant ventilation direction on shock wave propagation trends. Shock wave-ventilation collaborative propagation simulation data is constructed to reflect the dynamic interaction between the post-explosion pressure wave and the ventilation system. Peak concentration points, concentration abrupt change points, and boundary convergence points are identified in the spatial concentration residue map. These points are then combined into a temporally and spatially continuous sequence of explosion sign points, which is then output as a sequence of explosion characteristic point data. Based on the spatial distribution and temporal trend of the explosion characteristic point sequence, spatial vector trajectory fitting and backtracking methods are applied to construct potential backpropagation paths for explosion energy. This is then output as explosion energy backpropagation trajectory data for intersection analysis with co-propagation simulations. A three-dimensional spatial path intersection analysis is performed on the shock wave-ventilation co-propagation simulation data and the explosion energy backpropagation trajectory data. Areas with dense intersection points, areas with high probability of path overlap, and spatial segments with significant directional consistency are identified. Integrating spatial terrain, obstacle structures, and other information eliminates inaccessible or non-physically reasonable areas. Finally, the estimated coordinate domain of the explosion point—the inferred candidate explosion source area—is output, which can be used for emergency response or accident investigation.

[0042] Preferably, performing main vector ventilation guide identification on multi-directional equipotential vector tracking data includes: Perform vector direction consistency cluster analysis on multi-directional equipotential vector tracking data to generate equipotential vector main direction cluster data; evaluate the wind flow coupling intensity of equipotential vector main direction cluster data to generate ventilation adaptability matching score data; Perform multi-scale path topology mapping on the ventilation adaptability matching score data to generate ventilation path fitting grid data; perform time-series wind speed dynamic playback analysis on the ventilation path fitting grid data to generate dynamic ventilation flow trend map data; The dynamic ventilation flow trend map data is analyzed by combining path stability and dominance weight to generate ventilation dominant path characteristic data.

[0043] In an embodiment of the present invention, multi-directional equipotential vector tracking data is mapped to a three-dimensional vector field; vector directions are clustered according to consistency using a clustering algorithm based on direction cosines or spherical surfaces; vector groups with high directional stability and small fluctuations are preferentially retained in the clustering; and clustered equipotential vector main direction cluster data are output for subsequent wind flow adaptability analysis. The equipotential vector main direction clusters are spatially compared with the actual ventilation layout data of the mine; the consistency of each cluster direction with the air duct structure, wind pressure direction, and known flow velocity distribution is analyzed; the coupling strength of each direction cluster is scored using an adaptability scoring method; and ventilation adaptability matching score data is output as an important basis for screening dominant paths. Based on the adaptability matching score, directional clusters with high adaptability are screened; spatial projection and path reconstruction are performed at different scales (coarse-medium-fine); ventilation path fitting grid data reflecting the continuity and patency of the air flow channel are constructed; and node connection relationships with good dominant direction patency and geometric continuity are retained in the grid. Map existing wind speed observation data, simulation data, or time-series boundary conditions to a ventilation path fitting grid. Perform time-series progression analysis within the grid structure to simulate the propagation trend of wind speed changes over time. Output dynamic ventilation flow trend map data for evaluating path stability and dominance. Stability evaluation of the dynamic ventilation flow trend map is performed, including indicators such as wind speed fluctuation amplitude and direction retention rate. Simultaneously, a dominance weight analysis is performed, such as ventilation volume proportion and control capability of the path crossing area. The stability analysis results are weighted and fused with the dominance indicators to generate ventilation dominance path characteristic data reflecting the main wind flow channels, which is used for collaborative modeling with the shock wave propagation path.

[0044] Preferably, step S4 includes the following steps: Step S41: performing spatial fitting reconstruction on the residual gas concentration inversion network to generate high-dimensional inversion matrix data of gas concentration; Step S42: superimposing the explosion point estimated coordinate domain and the gas concentration high-dimensional inversion matrix data on the local explosion disturbance response to generate disturbance-enhanced gas concentration field data; Step S43: performing a three-dimensional multi-time series field fusion on the disturbance-enhanced gas concentration field data and the toxic gas spatial distribution evolution diagram to generate a dynamic evolution diagram of air quality after the coal mine disaster; Step S44: Use the preset regional concentration threshold to perform regional concentration threshold judgment on the dynamic evolution map of air quality after the coal mine disaster, and generate dangerous section concentration exceeding limit identification data; perform spatial distribution clustering and time series evolution trend analysis on the dangerous section concentration exceeding limit identification data, and generate a coal mine safety risk level annotation map to perform coal mine safety analysis operations.

[0045] In this embodiment of the present invention, a deep inversion neural network is constructed based on historical concentration sampling data, residual gas observation point data, and gas physical diffusion laws. The network inputs include spatial coordinates, gas type, wind speed and direction, temperature, and other characteristic dimensions. High-density sampling and fitting are performed on the entire coal mine space, outputting a high-dimensional fitted concentration matrix. Ultimately, a high-dimensional inversion matrix of gas concentration data is generated as the basic descriptive unit for post-disaster spatial concentration. The estimated coordinate domain of the explosion point obtained in step S33 is extracted and used as the center of the disturbance source. Local disturbance response modeling is applied to the high-dimensional inversion matrix data of gas concentration. The disturbance model takes into account the explosion intensity parameters, impact radius, and superimposed gas diffusion weighting factors. The disturbance-enhanced gas concentration field data with local pressure disturbance amplification characteristics is output. The disturbance-enhanced gas concentration field data is aligned with the toxic gas spatial distribution evolution map generated in step S34 in a three-dimensional time series. Differential evolution modeling is performed for multiple time series points (e.g., 5 minutes, 15 minutes, 30 minutes, and 1 hour). Gas types are classified and integrated, including toxic gases such as methane, carbon monoxide, and hydrogen sulfide. A visualization of post-disaster air quality is generated: a dynamic evolution map of post-disaster air quality, which is used for dynamic monitoring and decision support. Regional concentration exceedance standards are defined based on the company's specified safety thresholds for various gas concentrations. These concentration thresholds are mapped to the dynamic evolution map of coal mine air quality for region-by-region time series determination. Spatial location data for concentration anomaly sections, i.e., identification data for concentration exceedances in dangerous sections, is output. Spatial clustering (e.g., DBSCAN) and time series trend analysis (e.g., sliding window regression) are performed on the exceedance sections. Risk classification is performed based on parameters such as the degree of exceedance, duration, and spatial range. Finally, a coal mine safety risk level annotation map is generated for production scheduling and emergency command, providing an auxiliary decision-making basis for mine operation safety analysis.

[0046] In this specification, a coal mine safety analysis system based on big data analysis is provided, which is used to execute the above-mentioned coal mine safety analysis method based on big data analysis. The coal mine safety analysis system based on big data analysis includes: The gas field identification module is used to obtain coal mine engineering environmental data and coal mine engineering geographic data, where the coal mine engineering environmental data includes high-frequency gas concentration data and ventilation status data; extract the geographic topology information of the coal mine engineering geographic data, and integrate the coal mine engineering environmental data into the initial data set of the coal mine post-disaster gas field; The concentration analysis module is used to perform thermal-mass coupled diffusion analysis on the initial data set of the gas field after the coal mine disaster and generate a residual gas concentration inversion network. The residual gas concentration inversion network is used to invert the spatial distribution of residual gas on the initial data set of the gas field after the coal mine disaster and obtain the spatial concentration residual map of each gas component. The distribution evolution module is used to extract high-order concentration gradient boundary characteristic data from the spatial concentration residual map of each gas component, and perform reverse tracing calculation of the explosion starting point source on the high-order concentration gradient boundary characteristic data to generate the explosion point estimated coordinate domain; based on the explosion point estimated coordinate domain, the spatial concentration residual map is used to predict the spatial distribution evolution of the toxic gas residue to generate a toxic gas spatial distribution evolution map; The safety assessment module is used to couple and fuse the residual gas concentration inversion network, the explosion point estimation coordinate domain, and the toxic gas spatial distribution evolution map to construct a dynamic evolution map of air quality after a coal mine disaster; and to use preset regional concentration thresholds to conduct a safety risk assessment on the dynamic evolution map of air quality after a coal mine disaster to perform coal mine safety analysis operations.

[0047] The beneficial effect of the present invention is that the gas field identification module realizes the systematic collection and integration of complex environmental data after coal mine disasters by integrating high-frequency gas concentration data, ventilation status data and coal mine engineering geographical topology information, ensuring the comprehensiveness, timeliness and spatial accuracy of the initial data set of the gas field after the disaster, and providing a solid data foundation for subsequent diffusion analysis. The concentration analysis module relies on the thermal-mass coupling diffusion model to construct a residual gas concentration inversion network, which can accurately capture the multi-physical field diffusion characteristics of coal mine gas after the disaster, and realize high-precision inversion and component separation of the spatial concentration residual map of each gas component, effectively revealing the spatial distribution law of gas diffusion after the disaster. The distribution evolution module uses high-order concentration gradient boundary feature data, combined with the reverse tracing algorithm to complete the precise positioning of the explosion starting point, significantly improving the spatial accuracy of the explosion source estimation; based on the prediction of the spatial distribution evolution of toxic gas residues in the explosion point estimation coordinate domain, dynamic tracking and evolution simulation of the toxic gas diffusion process after the disaster are realized. The safety assessment module multi-dimensionally couples and fuses the residual gas concentration inversion network, the explosion point estimation coordinate domain, and the toxic gas spatial distribution evolution map to form a dynamic three-dimensional evolution map of post-disaster air quality, realizing real-time monitoring and trend prediction of the spatiotemporal changes of air pollution. By setting concentration thresholds for the dynamic evolution map to conduct risk assessment, rapid identification of dangerous areas and safety level division are achieved, providing scientific and quantitative decision support for coal mine safety management and emergency response, and significantly improving the efficiency and accuracy of coal mine post-disaster safety analysis. Therefore, the present invention realizes high-precision dynamic monitoring of the gas field after coal mine disasters and precise positioning of explosion sources through multi-source data fusion and thermal-mass coupling inversion, thereby improving the real-time assessment capability of coal mine safety risks.

[0048] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0049] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A coal mine safety analysis method based on big data analysis, characterized in that: The following steps are involved: Step S1: Acquire coal mine engineering environmental data and coal mine engineering geographic data, wherein the coal mine engineering environmental data includes high-frequency gas concentration data and ventilation status data; extract geographic topology information from the coal mine engineering geographic data, and integrate the coal mine engineering environmental data into an initial data set of coal mine post-disaster gas field; Step S2: performing a thermal-mass coupled diffusion analysis on the initial dataset of the gas field after the coal mine disaster to generate a residual gas concentration inversion network; using the residual gas concentration inversion network, performing a residual gas spatial distribution inversion on the initial dataset of the gas field after the coal mine disaster to obtain a spatial concentration residual map of each gas component; Step S3: extracting high-order concentration gradient boundary characteristic data of the spatial concentration residual map of each gas component, and performing reverse tracing calculation of the explosion starting point source on the high-order concentration gradient boundary characteristic data, thereby generating an explosion point estimated coordinate domain; based on the explosion point estimated coordinate domain, predicting the spatial distribution evolution of the toxic gas residue on the spatial concentration residual map, and generating a toxic gas spatial distribution evolution map; Step S4: Couple and fuse the residual gas concentration inversion network, explosion point estimation coordinate domain, and toxic gas spatial distribution evolution map to construct a dynamic evolution map of air quality after a coal mine disaster; use the preset regional concentration threshold to conduct a safety risk assessment on the dynamic evolution map of air quality after a coal mine disaster to perform coal mine safety analysis operations.

2. The coal mine safety analysis method based on big data analysis according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Acquire coal mine engineering environmental data and coal mine engineering geographical data, wherein the coal mine engineering environmental data includes high-frequency gas concentration data and ventilation status data; Step S12: resampling the high-frequency time series of the original coal mine engineering environment data to generate high-frequency gas concentration time data; Step S13: Divide the coal mine ventilation status data into regional ventilation modes to generate coal mine local ventilation topology data; Step S14: reconstructing the coal mine engineering geographic data into three-dimensional topology to generate coal mine geographic topology grid data; Step S15: Perform spatiotemporal fusion interpolation on the high-frequency gas concentration time data and the coal mine local ventilation topology structure data to generate coal mine ventilation-gas joint field data; integrate the coal mine ventilation-gas joint field data and the coal mine geographic topology grid data into the initial data set of the coal mine post-disaster gas field.

3. The coal mine safety analysis method based on big data analysis according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: using the heat-mass cooperative diffusion equation to estimate the heat-mass transfer coefficient of the initial data set of the gas field after the coal mine disaster; Step S22: performing dynamic diffusion field simulation on the initial data set of the coal mine post-disaster gas field through the heat-mass transfer coefficient to generate post-disaster gas residual diffusion evolution data; Step S23: performing residual fitting on the post-disaster gas residual diffusion evolution data to construct a residual gas concentration inversion network; Step S24: Use the residual gas concentration inversion network to perform spatial inversion calculation on the initial data set of the gas field after the coal mine disaster to generate three-dimensional reconstruction data of the residual gas concentration; perform multi-component spectrum separation and spatial cluster analysis on the three-dimensional reconstruction data of the residual gas concentration to generate a spatial concentration residual map of each gas component.

4. The coal mine safety analysis method based on big data analysis according to claim 3, characterized in that: The multi-component spectrum separation and spatial cluster analysis of the three-dimensional reconstructed residual gas concentration data in step S24 includes: Perform Fourier transform on the three-dimensional reconstructed data of residual gas concentration, and extract the spectrum response characteristics of CH4, CO, CO2, and H2S after the transformation to generate spectrum response characteristic data; Blind source separation is performed on the three-dimensional reconstruction data of residual gas concentration using spectrum response characteristic data to extract the response signals of independent gases such as CH4, CO, CO2, and H2S, and generate independent gas component concentration data sets. Perform three-dimensional spatial cluster analysis on independent gas component concentration data sets to generate gas component spatial clustering map data; Perform geometric fitting and boundary reconstruction on the gas component spatial clustering map data to generate spatial concentration isosurface data; The concentration data of independent gas components and the isosurface map data are integrated to generate the spatial concentration residual map of each gas component.

5. The coal mine safety analysis method based on big data analysis according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: calculating the local concentration change rate of each type of harmful gas in the spatial concentration residual map, and extracting high-order concentration gradient boundary feature data of the spatial concentration residual map based on the local concentration change rate; Step S32: performing toxic gas distribution analysis on each type of harmful gas in the spatial concentration residual map based on the local concentration change rate to construct a toxic gas distribution map; reconstructing the anti-diffusion vector field of the high-order concentration gradient boundary feature data to generate multi-directional equipotential vector tracking data; Step S33: performing a shock wave-ventilation path collaborative propagation path simulation on the high-order concentration gradient boundary characteristic data based on the multi-directional equipotential vector tracking data, and confirming the location of the initial energy source of the explosion by combining the simulation results with the concentration peak sequence and diffusion reverse vector trajectory of the spatial concentration residual map to generate an estimated coordinate domain of the explosion point; Step S34: Couple the explosion point estimated coordinate domain with the toxic gas distribution map to perform diffusion modeling to generate a time-varying initial diffusion field of the toxic gas; perform time-stepping simulation based on the time-varying initial diffusion field data of the toxic gas to generate a spatial distribution evolution diagram of the toxic gas residue.

6. The coal mine safety analysis method based on big data analysis according to claim 5, characterized in that: Step S33 includes the following steps: Step S331: performing regional shock wave response sensitivity division on the high-order concentration gradient boundary feature data to generate shock wave action weight distribution data; Step S332: performing main vector ventilation guide identification on the multi-directional equipotential vector tracking data to generate ventilation dominant path characteristic data; performing joint path mapping on the shock wave action weight distribution data and the ventilation dominant path characteristic data to generate shock wave-ventilation coordinated propagation simulation data; Step S333: extracting the concentration peak sequence of the spatial concentration residual map to generate explosion feature point sequence data; performing diffusion reverse vector fitting on the explosion feature point sequence data to generate explosion energy reverse propagation trajectory data; Step S334: performing path intersection analysis on the shock wave-ventilation cooperative propagation simulation data and the explosion energy reverse propagation trajectory data to generate an explosion point estimation coordinate domain.

7. The coal mine safety analysis method based on big data analysis according to claim 6, characterized in that: The main vector ventilation direction identification of multi-directional equipotential vector tracking data includes: Perform vector direction consistency cluster analysis on multi-directional equipotential vector tracking data to generate equipotential vector main direction cluster data; evaluate the wind flow coupling intensity of equipotential vector main direction cluster data to generate ventilation adaptability matching score data; Perform multi-scale path topology mapping on the ventilation adaptability matching score data to generate ventilation path fitting grid data; perform time-series wind speed dynamic playback analysis on the ventilation path fitting grid data to generate dynamic ventilation flow trend map data; The dynamic ventilation flow trend map data is analyzed by combining path stability and dominance weight to generate ventilation dominant path characteristic data.

8. The coal mine safety analysis method based on big data analysis according to claim 5, characterized in that: The time-stepping simulation based on the initial field data of the time-varying diffusion of poisonous gas in step S34 includes: Based on the time-varying diffusion initial field data of poisonous gas, multi-parameter heat and mass transport coefficients are extracted to generate diffusion evolution control parameter set data. The multi-parameter heat and mass transport coefficients include the mass diffusion coefficient, thermal diffusion coefficient and concentration gradient change rate of each gas. Initialize the three-dimensional multi-physics field time series grid of the diffusion evolution control parameter set data to generate the time-stepping cell data of the toxic gas diffusion; perform dynamic time series solution based on the steady state and boundary perturbations on the toxic gas diffusion time-stepping cell data to generate the multi-time concentration evolution field data of the toxic gas; Perform high-order spatial interpolation and continuity enhancement processing on the multi-time concentration evolution field data of poison gas to generate a time-continuous evolution map of poison gas concentration; The regional residual intensity cluster analysis is performed on the time-continuous evolution map of the toxic gas concentration to generate the spatial distribution evolution map of the toxic gas residue.

9. The coal mine safety analysis method based on big data analysis according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing spatial fitting reconstruction on the residual gas concentration inversion network to generate high-dimensional inversion matrix data of gas concentration; Step S42: superimposing the explosion point estimated coordinate domain and the gas concentration high-dimensional inversion matrix data on the local explosion disturbance response to generate disturbance-enhanced gas concentration field data; Step S43: performing a three-dimensional multi-time series field fusion on the disturbance-enhanced gas concentration field data and the toxic gas spatial distribution evolution diagram to generate a dynamic evolution diagram of air quality after the coal mine disaster; Step S44: Use the preset regional concentration threshold to perform regional concentration threshold judgment on the dynamic evolution map of air quality after the coal mine disaster, and generate dangerous section concentration exceeding limit identification data; perform spatial distribution clustering and time series evolution trend analysis on the dangerous section concentration exceeding limit identification data, and generate a coal mine safety risk level annotation map to perform coal mine safety analysis operations.

10. A coal mine safety analysis system based on big data analysis, characterized in that: For executing the coal mine safety analysis method based on big data analysis as claimed in claim 1, the coal mine safety analysis system based on big data analysis comprises: The gas field identification module is used to obtain coal mine engineering environmental data and coal mine engineering geographic data, where the coal mine engineering environmental data includes high-frequency gas concentration data and ventilation status data; extract the geographic topology information of the coal mine engineering geographic data, and integrate the coal mine engineering environmental data into the initial data set of the coal mine post-disaster gas field; The concentration analysis module is used to perform thermal-mass coupled diffusion analysis on the initial data set of the gas field after the coal mine disaster and generate a residual gas concentration inversion network. The residual gas concentration inversion network is used to invert the spatial distribution of residual gas on the initial data set of the gas field after the coal mine disaster and obtain the spatial concentration residual map of each gas component. The distribution evolution module is used to extract high-order concentration gradient boundary characteristic data from the spatial concentration residual map of each gas component, and perform reverse tracing calculation of the explosion starting point source on the high-order concentration gradient boundary characteristic data to generate the explosion point estimated coordinate domain; based on the explosion point estimated coordinate domain, the spatial concentration residual map is used to predict the spatial distribution evolution of the toxic gas residue to generate a toxic gas spatial distribution evolution map; The safety assessment module is used to couple and fuse the residual gas concentration inversion network, the explosion point estimation coordinate domain, and the toxic gas spatial distribution evolution map to construct a dynamic evolution map of air quality after a coal mine disaster; and to use preset regional concentration thresholds to conduct a safety risk assessment on the dynamic evolution map of air quality after a coal mine disaster to perform coal mine safety analysis operations.

Images