Coalfield aquifer feature identification method

Abstract

The application relates to the technical field of geological exploration, and discloses a coalfield aquifer feature identification method. The coalfield aquifer feature identification method is characterized in that: a multi-source data fusion modeling is adopted to realize fine reconstruction of a three-dimensional heterogeneous structure of a structural belt; a permeability tensor field construction is adopted to realize a leap from a scalar fuzzy description to a tensor accurate description of water conductivity; a fault gouge shear slip and a permeability coefficient increment are quantitatively associated by coupling calculation of mining stress to predict an activation risk; an aquifer boundary and a permeability parameter are corrected in real time and synchronously along with a mining process by dynamic updating identification; a model self-evolution and identification precision continuous iteration improvement capability is formed by closed-loop verification optimization; the control water effect of the structural belt is improved from a traditional binary coarse division of water conduction and water isolation to a three-dimensional heterogeneous fine description; a water inrush risk prediction is changed from a static lagging evaluation to a real-time early warning; and the aquifer feature identification is dynamically synchronized with the actual mining in a whole mining period.

Description

Technical Field

[0001] This application relates to the field of geological exploration technology, and in particular to a method for identifying the characteristics of coalfield aquifers. Background Technology

[0002] The field of coalfield hydrogeological exploration and mine water hazard prevention technology involves the identification, evaluation, and management of groundwater systems during coal seam mining. With the increasing depth of coal mining, water inrush disasters from confined aquifers in the coal seam floor have become one of the main types of hazards threatening mine safety. Accurately identifying the spatial distribution, water-bearing characteristics, and spatial relationship between coalfield aquifers and mining operations is a fundamental prerequisite for conducting water inrush hazard assessments and formulating water prevention measures. In aquifer characteristic identification technology, the refined characterization of structural water-controlling effects is a core element. Faults, collapse columns, and other structural zones, as key elements controlling the hydraulic connection of aquifers, directly determine the accuracy of aquifer identification and the effectiveness of water hazard prediction due to their internal structural zonation, anisotropic conductivity, and mining-induced activation effects.

[0003] Existing technologies for identifying coalfield aquifer characteristics have shortcomings in characterizing the structural water-controlling effect. Specifically, existing technologies can only obtain one-dimensional zoning information of structural zones through borehole core description, with a spatial sampling rate of less than 0.1%. Furthermore, the Kriging interpolation method based on sparse boreholes assumes spatial stationarity, which contradicts the inherent characteristics of non-stationary random fields within structural zones. This leads to significant uncertainties in the three-dimensional heterogeneous characterization of structural zones between boreholes. Simultaneously, the permeability coefficients obtained from existing pumping tests are equivalent isotropic parameters, failing to reflect the tensor characteristics of fault zones along strike, perpendicular to strike, and vertical. Multi-well pumping tests can only characterize anisotropy at the local borehole scale and cannot be extrapolated to the overall fault zone scale, severely limiting the accuracy of identifying water-conducting channels within structural zones. Moreover, existing technologies lack quantitative models of the evolution of mining stress-fault gouge dilatation-permeability, making it impossible to predict under what mining conditions a previously water-blocking fault will transform into a water-conducting channel. This results in a severe disconnect between aquifer identification results and actual mining activities, increasing the risk of inaccurate predictions of water inrush disasters. Summary of the Invention

[0004] This application proposes a method for identifying coalfield aquifer characteristics. This method possesses advantages such as multi-source data fusion modeling, permeability tensor field construction, mining-induced stress coupling calculation, dynamic update identification, and closed-loop verification and optimization. It addresses the problems of existing technologies, which are constrained by insufficient control of sparse boreholes and the assumption of equivalent isotropy. These limitations lead to significant uncertainties in the characterization of the three-dimensional heterogeneous structure of the geological zone, a lack of quantitative methods for water conductivity anisotropy, difficulty in predicting aquifer boundary abrupt changes due to fault activation under mining conditions, a disconnect between static models and actual mining activities, and the accumulation of errors in a unidirectional linear process. This method forms a closed-loop architecture through five progressive steps. The preceding steps provide the data and model foundation for subsequent steps, and the subsequent steps dynamically correct the results of the preceding steps and feed them back to the knowledge base, ensuring that the aquifer identification results remain dynamically synchronized with actual mining activities throughout the entire mining cycle.

[0005] To achieve the above objectives, this application adopts the following technical solution: a method for identifying the characteristics of coalfield aquifers, characterized by the following specific steps: S1. Multi-source fusion modeling: Collect multi-source data, establish a spatial framework, identify fault zone boundaries and lithological zoning, and generate a three-dimensional structural model; S2. Construction of permeability tensor: Based on the above model, permeability coefficient tensors are assigned to lithological zoning to establish a permeability tensor field, identify water-conducting channels and water-impermeable barriers, and output anisotropic distribution map; S3. Mining-induced stress coupling: Establish a numerical model for coalfield mining, input coal seam occurrence conditions and overlying mechanical parameters, calculate the redistribution of the three-dimensional stress field caused by mining, extract the stress tensor of the structural zone, determine the activation risk, and output the coupled response field of mining-induced stress and structural zone. S4. Dynamic Update Identification: The activated region is used as a dynamic boundary condition to drive the tensor field update and correct the permeability tensor, establish the mining-driven seepage coupling equation, inject monitoring data to invert aquifer parameters, and output dynamic identification results. S5. Closed-loop verification and optimization: The identification results are compared with the water inflow point and water quality data to calculate the consistency index. Based on the deviation, the model self-optimization is triggered by reverse tracing, and the optimization parameters are fed back to form a closed-loop self-evolution system.

[0006] Furthermore, in step S1, the specific steps of multi-source fusion modeling are as follows: S1.1 Multi-source data spatial registration: Collect three-dimensional seismic, transient electromagnetic and hydrological borehole data in the coalfield area, establish a unified spatial framework through time-depth conversion and coordinate registration, identify fault zone boundaries based on seismic wave impedance differences, delineate the influence range of tectonic zones in combination with transient electromagnetic low-resistivity anomalies, and mark the lithological zoning interfaces within the tectonic zones based on borehole cores. S1.2 Generation of three-dimensional heterogeneous structure: The registered multi-source data is input into the geostatistical stochastic simulation framework. The spatial correlation structure is established through variogram analysis under the constraint of lithological zoning, generating a heterogeneous structure model of tectonic zone. The three-dimensional distribution model of fault core, fracture zone and fracture influence zone is output and used as the input data for step S2.

[0007] Furthermore, in step S2, the specific steps for constructing the permeation tensor are as follows: S2.1 Zonal Tensor Assignment: Based on the three-dimensional structural model output in step S1, the fault core, fractured rock zone and fracture influence zone are assigned corresponding permeability tensors according to the lithological zoning characteristics of the tectonic zone. The principal direction of the permeability tensor of the fracture influence zone is calculated based on the fracture orientation data. The fracture network equivalence method is used to transform the discrete fracture distribution into the permeability tensor of the equivalent continuous medium. S2.2 Anisotropic Field Identification: Based on the zonal assignment results, a full-space permeability tensor field of the structural zone is established. Through the analysis of the principal directions and principal values ​​of the tensor, high water-conducting channels along the fault strike and water-blocking barriers perpendicular to the strike are identified. The anisotropic distribution map of water conductivity of the structural zone is output and used as input data for the stress coupling calculation of mining in step S3.

[0008] Furthermore, in step S2.1, the specific steps for assigning values ​​to the segmented tensor are as follows: S2.11 Initialization of Lithological Zoning Tensors: Based on the three-dimensional structural model in step S1, permeability coefficient tensors of fault cores and fractured rock zones are initialized according to the lithological zoning characteristics of the tectonic zone, and permeability tensor fields of the tectonic zone matrix are constructed. The principal value of the fault core tensor corresponds to low permeability characteristics, and the principal value of the fractured rock zone tensor corresponds to medium permeability characteristics. S2.12, Superposition of equivalent fracture tensors: Extract fracture orientation data of the fracture influence zone, calculate the normal and tangential permeability coefficients of the fracture surface, and use the fracture network equivalence method to transform the discrete fracture distribution into an equivalent continuous medium permeability tensor, which is superimposed on the matrix skeleton tensor field to form a heterogeneous anisotropic permeability tensor field of the entire tectonic zone.

[0009] Furthermore, in step S2.2, the specific steps for anisotropic field identification are as follows: S2.21 Tensor Feature Analysis: Based on the full-space permeability tensor field of the tectonic zone established in step S2.1, feature decomposition is performed on the tensors of each spatial node to extract the principal direction and principal value. High water-conducting channels with dominant fracture connectivity are identified along the fault strike, and water-impermeable barriers formed by fault gouge are identified perpendicular to the fault strike, thus characterizing the three-dimensional anisotropic water-conducting pattern of the tectonic zone. S2.22 Anisotropic Mapping: Based on the tensor feature analysis results, generate anisotropic distribution map of water conductivity in the structural zone, identify the spatial location of water-conducting channels, the distribution range of water barriers, and the permeability tensor parameters of nodes, and register the distribution map with the mining stress field in spatial coordinates to determine the background conditions of permeability in the structural zone and use it as input data for the mining stress coupling calculation in step S3.

[0010] Furthermore, in step S3, the specific steps of stress coupling are as follows: S3.1 Mining-induced stress analysis: Establish a numerical model for coalfield mining, input coal seam occurrence conditions, mining methods and overlying strata mechanical parameters, calculate the redistribution of the three-dimensional stress field caused by mining, extract the stress tensor of the structural zone, analyze the shear stress, normal stress and shear stress ratio of the fault zone, determine the activation state of different parts of the structural zone according to the fault gouge shear slip criterion, and extract the critical shear stress ratio threshold. S3.2 Activation Region Mapping: Map the mining stress analysis results to the permeability tensor field established in step S2, mark the spatial location and geometry of potential activation regions, record the ratio of shear stress to normal stress and slip displacement in the activation regions, and construct the coupled response field of mining stress and structural zone as the driving data for permeability tensor field correction and parameter inversion in step S4 dynamic update identification.

[0011] Furthermore, in step S4, the specific steps for dynamically updating the identification are as follows: S4.1 Tensor Dynamic Reconstruction: The activated region in step S3 is used as a dynamic boundary condition to drive the reconstruction and update of the permeability tensor field in step S2. The permeability tensor of the fault gouge in the activated region is corrected according to the relationship between shear slip and permeability coefficient increment. The coupled equation of mining and seepage is established to solve the flow field evolution. S4.2 Multi-source data assimilation: Water level monitoring and microseismic event data are injected into the mining and seepage coupling model through data assimilation algorithm. The aquifer boundary and seepage parameters are jointly inverted and corrected, and the dynamically updated aquifer feature identification results are output as the benchmark data for verification and optimization in step S5.

[0012] Furthermore, in step S4.1, the specific steps for tensor dynamic reconstruction are as follows: S4.11 Activation Boundary Drive: The spatial coordinates and geometric shape of the activation region output in step S3 are used as dynamic boundary conditions to drive the reconstruction and update of the permeability tensor field established in step S2, determine the boundary node coordinates of the activation region and the initial value of the permeability tensor, and establish the mapping relationship between the dynamic boundary conditions and the static tensor field. S4.12, Fault gouge tensor correction: Based on the relationship model between shear slip and permeability coefficient increment in the activated region, the original low permeability tensor of the fault gouge in the activated region is corrected and a high permeability tensor is assigned. The coupled control equations of mining and seepage are established, and the evolution of the aquifer flow field is solved using the corrected tensor field as the parameter field. The pressure distribution and velocity vector field of the flow field are output.

[0013] Furthermore, in step S4.2, the specific steps for multi-source data assimilation are as follows: S4.21 Monitoring Data Injection: Water level monitoring data and microseismic event data are used as observation datasets. Observation operators are constructed to establish the mapping relationship between monitoring data and model state variables. The observation dataset is integrated into the mining-induced and seepage coupling model through a data assimilation algorithm. The model state estimates are iteratively updated to reduce the deviation between model predictions and measured data. S4.22 Boundary Parameter Inversion: Based on the updated model state estimate, the boundary morphology and permeability tensor parameters of the aquifer are jointly inverted and corrected to generate dynamically updated aquifer feature identification results. These results include the spatial distribution of the aquifer, the location of the water-conducting channels, and the distribution of permeability parameters, and serve as the benchmark data for the closed-loop verification and optimization step S5.

[0014] Furthermore, in step S5, the specific steps for closed-loop verification optimization are as follows: S5.1 Multi-scale verification: The dynamic aquifer feature identification results of step S4 are compared and verified with the actual water inflow points and water quality monitoring data in the well at multiple scales. The consistency index between the identification results and the actual water inflow points in three dimensions—spatial location, water inflow volume, and water chemistry type—is calculated to quantify the spatial distribution characteristics and magnitude of the identification deviation. S5.2 Deviation Source Tracing and Optimization: Based on the verification deviation, trace back to the structural model in step S1, the tensor parameters in step S2, or the stress calculation results in step S3 to locate the source of the deviation and trigger the self-optimization of the corresponding step model. Write the optimized parameters and structure into the knowledge base and feed them back to step S1 for the next round of identification, forming a closed-loop self-evolutionary system.

[0015] The beneficial effects of this invention are as follows: This application provides a method for identifying coalfield aquifer characteristics. It achieves fine reconstruction of the three-dimensional heterogeneous structure of the structural zone through multi-source data fusion modeling, and realizes the leap from scalar fuzzy description to tensor precise characterization of water conductivity through the construction of a permeability tensor field. It establishes a quantitative correlation between fault gouge shear slip and permeability coefficient increment through mining-induced stress coupling calculation to predict activation risk. It realizes real-time synchronous correction of aquifer boundary and permeability parameters with mining process through dynamic update identification, and forms the ability of model self-evolution and continuous iterative improvement of identification accuracy through closed-loop verification and optimization. It realizes the effect of improving the water control effect of the structural zone from the traditional binary coarse division of water-conducting and water-blocking to a three-dimensional heterogeneous fine characterization, changing the prediction of water inrush risk from static lag assessment to real-time advanced early warning, and keeping the aquifer characteristic identification dynamically synchronized with the actual mining operation throughout the entire mining cycle. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort: Figure 1 This is a diagram illustrating the overall method steps of this application.

[0017] Figure 2 This is an extended step diagram of step S2 in this application.

[0018] Figure 3 This is an extended step diagram of step S2.1 of this application.

[0019] Figure 4 This is an extended step diagram of step S2.2 of this application.

[0020] Figure 5 This is an extended step diagram for step S4 of this application.

[0021] Figure 6 This is an extended step diagram of step S4.1 of this application.

[0022] Figure 7 This is an extended step diagram of step S4.2 of this application. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Example 1, as Figure 1 A method for identifying the characteristics of coalfield aquifers, characterized by the following specific steps: S1. Multi-source fusion modeling: Collect multi-source data, establish a spatial framework, identify fault zone boundaries and lithological zoning, and generate a three-dimensional structural model; S2. Construction of permeability tensor: Based on the above model, permeability coefficient tensors are assigned to lithological zoning to establish a permeability tensor field, identify water-conducting channels and water-impermeable barriers, and output anisotropic distribution map; S3. Mining-induced stress coupling: Establish a numerical model for coalfield mining, input coal seam occurrence conditions and overlying mechanical parameters, calculate the redistribution of the three-dimensional stress field caused by mining, extract the stress tensor of the structural zone, determine the activation risk, and output the coupled response field of mining-induced stress and structural zone. S4. Dynamic Update Identification: The activated region is used as a dynamic boundary condition to drive the tensor field update and correct the permeability tensor, establish the mining-driven seepage coupling equation, inject monitoring data to invert aquifer parameters, and output dynamic identification results. S5. Closed-loop verification and optimization: The identification results are compared with the water inflow point and water quality data to calculate the consistency index. Based on the deviation, the model self-optimization is triggered by reverse tracing, and the optimization parameters are fed back to form a closed-loop self-evolution system.

[0025] In this embodiment: Step S1 involves collecting 3D seismic, transient electromagnetic, and hydrogeological borehole data from the coalfield area, establishing a unified spatial framework through time-depth conversion and coordinate registration, identifying the macroscopic boundaries of fault zones based on seismic wave impedance differences, delineating the influence range of tectonic zones by combining transient electromagnetic low-resistivity anomalies, calibrating the internal lithological zoning interfaces using borehole cores, and inputting multi-source data into a geostatistical stochastic simulation framework to generate a 3D heterogeneous structural model of the tectonic zone. This completes the task of finely depicting the spatial distribution of fault cores, fractured rock zones, and fracture influence zones, overcoming the limitations of insufficient spatial sampling rate caused by sparse borehole control in existing technologies, achieving the goal of high-precision characterization of the 3D heterogeneous structure of the tectonic zone, and providing a reliable spatial structural foundation for the subsequent construction of the permeability tensor field.

[0026] Step S2, based on the three-dimensional structural model output from step S1, assigns corresponding permeability tensors to the fault core, fractured rock zone, and fracture-affected zone, respectively. For the fracture-affected zone, the principal direction of the permeability tensor is calculated based on the fracture attitude data. The fracture network equivalence method is used to transform discrete fractures into equivalent continuous medium permeability tensors, establishing a full-space permeability tensor field for the structural zone. Through the analysis of the principal direction and principal value of the tensor, water-conducting channels and water-blocking barriers are identified, completing the task of accurately quantifying the anisotropy of the hydraulic conductivity of the structural zone. This overcomes the shortcomings of existing technologies that use equivalent isotropic parameters to reflect the tensor characteristics of the fault zone, achieving the goal of transitioning from scalar fuzzy description of hydraulic conductivity to precise tensor characterization, and objectively representing the strength and directional characteristics of hydraulic conductivity in different spatial parts of the structural zone.

[0027] Step S3 establishes a numerical model for coalfield mining, inputs coal seam occurrence conditions and overlying strata mechanical parameters, calculates the redistribution of the three-dimensional stress field caused by mining, extracts the stress tensor of the structural zone and calculates shear stress, normal stress and shear stress ratio, determines the activation risk of different parts of the structural zone based on the fault gouge shear slip criterion, maps the mechanical response results to the permeability tensor field and marks potential activation areas, and completes the task of quantitatively correlating mining stress with the mechanical response of the structural zone. This overcomes the limitation of existing technologies that lack quantitative relationship between mining stress and fault gouge permeability evolution, achieves the purpose of advanced prediction of the activation state of the structural zone, and provides mechanically driven boundary conditions for dynamic update identification.

[0028] Step S4 uses the activated region output from step S3 as a dynamic boundary condition to drive the reconstruction and update of the permeability tensor field. Based on the relationship between shear slip and permeability coefficient increment, the permeability tensor of the fault gouge in the activated region is corrected. The coupled control equation of mining and seepage is established to solve the evolution of the aquifer flow field. The real-time monitoring data is injected into the coupled model through the data assimilation algorithm to invert and correct the aquifer boundary and parameters. The task of dynamic synchronous correction of aquifer characteristics is completed, which breaks through the limitation of the existing static model being seriously out of touch with the actual mining. It achieves the goal of real-time evolution of the identification results with the mining process, and transforms the assessment of water inrush risk from lagging analysis to real-time early warning.

[0029] Step S5 compares and verifies the dynamic identification results output in step S4 with the actual water inflow points and water quality monitoring data in the well at multiple scales, calculates the consistency index of spatial location, water inflow level and water chemical type, traces back to the structural model, tensor parameters or stress calculation results based on the verification deviation, and triggers the self-optimization of the corresponding step model. The optimized parameters and structure are written into the knowledge base and fed back to step S1, completing the task of continuous iterative improvement of identification accuracy. It breaks through the limitation of the unidirectional linear process error accumulation and transmission of the existing technology, achieves the purpose of model self-evolution and closed-loop optimization of identification accuracy, and forms a self-evolving system that continuously improves with the increase of application frequency.

[0030] Existing technologies are constrained by insufficient control density of sparse boreholes and the assumption of equivalent isotropy. The characterization of the three-dimensional heterogeneous structure inside the tectonic zone is uncertain, there is a lack of effective means to quantify the anisotropy of water conductivity, and it is difficult to predict abrupt changes in aquifer boundaries caused by fault activation under mining conditions. This application uses multi-source data fusion modeling as a basis to construct a three-dimensional heterogeneous structural model of fault core, fractured rock zone, and fracture influence zone, solving the problem of fine reconstruction of the spatial structure of the tectonic zone. Then, it establishes a permeability tensor field of the entire space of the tectonic zone, representing the differences in water conductivity of different lithological zones in tensor form, overcoming the limitation that scalar parameters cannot reflect anisotropy. On this basis, it introduces mining-induced stress coupling calculation to establish a quantitative correlation between fault gouge shear slip and permeability coefficient increment, predicting potential activation areas of the tectonic zone. Subsequently, the activation area is used as a dynamic boundary condition to drive the reconstruction and update of the tensor field. Combined with data assimilation algorithm to inject real-time monitoring data, the aquifer boundary and permeability parameters are continuously corrected during the mining process. Finally, through closed-loop verification and optimization, the identification results are compared with the measured water inflow data in the well, the source of deviation is traced back and the model self-correction is triggered. This application forms a progressive closed-loop architecture through the above five steps. The preceding steps provide the data and model foundation for the subsequent steps, and the subsequent steps dynamically correct the results of the preceding steps and feed them back to the knowledge base. This ensures that the aquifer identification results remain dynamically synchronized with the actual mining activities throughout the entire mining cycle. The water control effect of the structure zone is improved from the traditional binary rough division of water-conducting and water-blocking to a three-dimensional heterogeneous fine characterization. The prediction of water inrush risk is transformed from static lag assessment to real-time advanced early warning, which improves the scientificity and reliability of coalfield water hazard prevention and control.

[0031] Example 2, as Figure 1 In step S1, the specific steps of multi-source fusion modeling are as follows: S1.1 Multi-source data spatial registration: Collect three-dimensional seismic, transient electromagnetic and hydrological borehole data in the coalfield area, establish a unified spatial framework through time-depth conversion and coordinate registration, identify fault zone boundaries based on seismic wave impedance differences, delineate the influence range of tectonic zones in combination with transient electromagnetic low-resistivity anomalies, and mark the lithological zoning interfaces within the tectonic zones based on borehole cores. S1.2 Generation of three-dimensional heterogeneous structure: The registered multi-source data is input into the geostatistical stochastic simulation framework. The spatial correlation structure is established through variogram analysis under the constraint of lithological zoning, generating a heterogeneous structure model of tectonic zone. The three-dimensional distribution model of fault core, fracture zone and fracture influence zone is output and used as the input data for step S2.

[0032] In this embodiment: Step S1.1 involves collecting 3D seismic, transient electromagnetic, and hydrogeological borehole data from the coalfield area, and establishing a unified spatial framework through time-depth transformation and coordinate registration. The macroscopic boundaries of fault zones are identified based on seismic wave impedance differences, and the influence range of tectonic zones is delineated using transient electromagnetic low-resistivity anomalies. Simultaneously, borehole core description data is used to calibrate the lithological zoning interfaces within the tectonic zones. This completes the accurate registration of multi-source heterogeneous data in the spatial and depth domains, fusing discrete point borehole data with continuous geophysical field data into a structured dataset under a unified coordinate system. This eliminates the scale mismatch and spatial misalignment problems of multi-source data, achieving the goal of providing basic data support for 3D modeling of tectonic zones.

[0033] Step S1.2 inputs the registered multi-source data into a geostatistical stochastic simulation framework, constructs a variogram function and establishes a spatial correlation structure under lithological zoning constraints. Using the lithological zoning interfaces calibrated from borehole cores as hard constraints, stochastic simulations are performed on the tectonic zone boundaries interpreted from seismic and electromagnetic data to generate a three-dimensional heterogeneous structural model of the tectonic zone. This completes the task of three-dimensional heterogeneous characterization of the internal structure of the tectonic zone, overcoming the limitations of one-dimensional point-like description caused by insufficient spatial sampling rate of sparse boreholes. This achieves a fine reconstruction of the three-dimensional spatial distribution of lithological zoning within the tectonic zone, providing a reliable spatial structural foundation for the construction of the permeability tensor field in step S2.

[0034] Steps S1.1 and S1.2 together constitute the multi-source data fusion modeling process in step S1. Existing technologies are limited by excessively large borehole spacing and insufficient spatial sampling rate of structural zones. They rely on sparse borehole interpolation to build geological models, making it difficult to reflect the spatial variation characteristics of heterogeneous structures within fault zones. Step S1 first fuses seismic, electromagnetic, and borehole data into a unified framework through multi-source data spatial registration, and then generates a three-dimensional heterogeneous structural model using geostatistical random simulation constrained by lithological zoning. This process allows for a detailed depiction of the spatial distribution of fault cores, breccia zones, and fracture influence zones within structural zones, elevating structural identification from a one-dimensional point-like description relying on sparse boreholes to a three-dimensional volumetric representation incorporating multi-source data. This reduces the uncertainty of inter-borehole interpolation, laying a reliable spatial structural foundation for subsequent permeability tensor field construction and hydrodynamic anisotropy analysis.

[0035] Example 3, as Figure 2 In step S2, the specific steps for constructing the permeation tensor are as follows: S2.1 Zonal Tensor Assignment: Based on the three-dimensional structural model output in step S1, the fault core, fractured rock zone and fracture influence zone are assigned corresponding permeability tensors according to the lithological zoning characteristics of the tectonic zone. The principal direction of the permeability tensor of the fracture influence zone is calculated based on the fracture orientation data. The fracture network equivalence method is used to transform the discrete fracture distribution into the permeability tensor of the equivalent continuous medium. S2.2 Anisotropic Field Identification: Based on the zonal assignment results, a full-space permeability tensor field of the structural zone is established. Through the analysis of the principal directions and principal values ​​of the tensor, high water-conducting channels along the fault strike and water-blocking barriers perpendicular to the strike are identified. The anisotropic distribution map of water conductivity of the structural zone is output and used as input data for the stress coupling calculation of mining in step S3.

[0036] In this embodiment: Step S2.1, based on the three-dimensional structural model established in Step S1, initializes the permeability tensors of the fault core, fractured rock zone, and fracture-affected zone according to the lithological zoning characteristics of the tectonic zone. This step calculates the principal direction of the permeability tensor for the fracture-affected zone based on fracture attitude data, and uses a fracture network equivalence method to transform the discrete fracture distribution into an equivalent continuous medium permeability tensor. This establishes a quantitative characterization system for the permeability characteristics of different lithological zoning within the tectonic zone, completing the task of converting discrete fractures in the fracture-affected zone into an equivalent continuous medium, and achieving the goal of providing a zoning parameter basis for constructing the full-space permeability tensor field of the tectonic zone.

[0037] Step S2.2 establishes the full-space permeability tensor field of the structural zone based on the zonal assignment results. This step identifies high water-conducting channels along the fault strike and water-retaining barriers perpendicular to the strike through tensor principal direction and principal value analysis, generating anisotropic distribution maps of water conductivity in the structural zone. This accurately characterizes the strength and direction of water conductivity in different spatial locations of the structural zone, completing the task of spatially locating water-conducting channels and water-retaining barriers, and achieving the goal of providing permeability background conditions for the mining-induced stress coupling calculation in Step S3.

[0038] Steps S2.1 and S2.2 together constitute the permeability tensor field construction process in step S2. Existing technologies use pumping tests to obtain equivalent isotropic permeability coefficients, which cannot reflect the differences in hydraulic conductivity along the strike, perpendicular to the strike, and vertically of the fault zone. Furthermore, multi-well pumping tests can only characterize anisotropy at the local borehole scale, making it difficult to extrapolate to the overall structural zone scale. Step S2 first initializes the permeability coefficient tensor for different parts of the structural zone based on lithological zoning characteristics. Then, it transforms the discrete fractures in the fracture-affected zone into equivalent continuous medium permeability tensors, ultimately establishing a full-space permeability tensor field for the structural zone and identifying water-conducting channels and water-retaining barriers. This process elevates the evaluation of structural zone hydraulic conductivity from a scalar-based fuzzy description to a precise tensor characterization, overcoming the limitations of the equivalent isotropic assumption. It extends the identification of hydraulic anisotropy from the local borehole scale to the overall structural zone scale, improving the objectivity and accuracy of the structural zone's water-controlling effect characterization. This lays the foundation for permeability parameters for subsequent mining-induced stress coupling calculations and activation zone identification.

[0039] Example 4, as Figure 3 In step S2.1, the specific steps for assigning values ​​to the segmented tensor are as follows: S2.11 Initialization of Lithological Zoning Tensors: Based on the three-dimensional structural model in step S1, permeability coefficient tensors of fault cores and fractured rock zones are initialized according to the lithological zoning characteristics of the tectonic zone, and permeability tensor fields of the tectonic zone matrix are constructed. The principal value of the fault core tensor corresponds to low permeability characteristics, and the principal value of the fractured rock zone tensor corresponds to medium permeability characteristics. S2.12, Superposition of equivalent fracture tensors: Extract fracture orientation data of the fracture influence zone, calculate the normal and tangential permeability coefficients of the fracture surface, and use the fracture network equivalence method to transform the discrete fracture distribution into an equivalent continuous medium permeability tensor, which is superimposed on the matrix skeleton tensor field to form a heterogeneous anisotropic permeability tensor field of the entire tectonic zone.

[0040] In this embodiment: Step S2.11, based on the three-dimensional structural model established in Step S1, initializes the permeability tensors of the fault core and the fractured rock zone according to the lithological zoning characteristics of the tectonic zone. This step sets the principal value of the fault core tensor to low permeability and the principal value of the fractured rock zone tensor to medium permeability, thereby constructing the permeability tensor field of the tectonic zone matrix framework. This process completes the task of quantitatively characterizing the permeability characteristics of different lithological zoning within the tectonic zone, overcoming the limitation of existing technologies that cannot distinguish the permeability differences between the fault core and the fractured rock zone using equivalent isotropic parameters, and achieving the goal of providing a matrix framework for the superposition of equivalent fracture tensors.

[0041] Step S2.12 extracts fracture attitude data from the fracture-affected zone, calculates the normal and tangential permeability coefficients of the fracture surface, and uses the fracture network equivalence method to transform the discrete fracture distribution into an equivalent continuous medium permeability tensor, which is then superimposed onto the matrix framework tensor field. This process completes the task of transforming discrete fractures in the fracture-affected zone into an equivalent continuous medium, solves the technical problem of the difficulty in comprehensively characterizing the permeability of fractured rock masses, and achieves the goal of forming a full-space heterogeneous anisotropic permeability tensor field of the tectonic zone.

[0042] Steps S2.11 and S2.12 together construct the zonal tensor assignment process in step S2.1. Existing technologies obtain equivalent isotropic permeability coefficients through pumping tests, which cannot reflect the difference in water conductivity along and perpendicular to the strike of the fault zone, and can only characterize the local-scale characteristics of the borehole. Step S2.1 first initializes the permeability coefficient tensor of the fault core and fractured rock zone based on lithological zoning characteristics, constructing a matrix framework tensor field. Then, the discrete fractures in the fracture-affected zone are transformed into equivalent continuous medium permeability tensors and superimposed onto the matrix framework, forming a heterogeneous anisotropic permeability tensor field of the entire tectonic zone. This process quantitatively distinguishes the permeability characteristics of different lithological zones within the tectonic zone, quantifies the overall permeability contribution of discrete fractures in the fracture-affected zone, and elevates the evaluation of the water conductivity of the tectonic zone from a scalar fuzzy description to a precise tensor characterization, improving the objectivity and accuracy of the characterization of the water control effect of the tectonic zone.

[0043] Example 5, such as Figure 4 In step S2.2, the specific steps for anisotropic field identification are as follows: S2.21 Tensor Feature Analysis: Based on the full-space permeability tensor field of the tectonic zone established in step S2.1, feature decomposition is performed on the tensors of each spatial node to extract the principal direction and principal value. High water-conducting channels with dominant fracture connectivity are identified along the fault strike, and water-impermeable barriers formed by fault gouge are identified perpendicular to the fault strike, thus characterizing the three-dimensional anisotropic water-conducting pattern of the tectonic zone. S2.22 Anisotropic Mapping: Based on the tensor feature analysis results, generate anisotropic distribution map of water conductivity in the structural zone, identify the spatial location of water-conducting channels, the distribution range of water barriers, and the permeability tensor parameters of nodes, and register the distribution map with the mining stress field in spatial coordinates to determine the background conditions of permeability in the structural zone and use it as input data for the mining stress coupling calculation in step S3.

[0044] In this embodiment: Step S2.21, based on the full-space permeability tensor field of the structural zone established in step S2.1, performs feature decomposition on the tensor of each spatial node and extracts the principal direction and principal value. Highly conductive channels with dominant fracture connectivity are identified along the fault strike, and water-impermeable barriers formed by fault gouge are identified perpendicular to the fault strike. From this point, the three-dimensional anisotropic water-conducting pattern of the structural zone is drawn. This process completes the task of quantitatively analyzing the strength and direction characteristics of water conductivity in different spatial parts of the structural zone, overcoming the limitation of existing technologies that cannot distinguish water-conducting directions using equivalent isotropic parameters, and achieving the goal of accurately locating the spatial distribution of water-conducting channels and water-impermeable barriers.

[0045] Step S2.22 generates an anisotropic distribution map of the hydraulic conductivity of the structural zone based on the tensor feature analysis results, identifying the spatial location of the water-conducting channels, the distribution range of the water-blocking barriers, and the nodal permeability tensor parameters. This distribution map is then spatially registered with the mining-induced stress field to determine the background conditions of permeability in the structural zone. This process completes the task of visualizing and spatially locating the hydraulic conductivity of the structural zone, solving the technical problem that existing technologies cannot spatially correlate permeability parameters with the mining-induced stress field, and achieving the goal of providing background permeability data for the mining-induced stress coupling calculation in step S3.

[0046] Steps S2.21 and S2.22 together construct the anisotropic field identification in step S2. Existing technologies use pumping tests to obtain equivalent isotropic permeability coefficients, which cannot reflect the difference in water conductivity along the strike and perpendicular to the fault zone. Furthermore, multi-hole pumping tests can only characterize anisotropy at the local borehole scale and are difficult to extrapolate to the overall structural zone scale. Step S2 first performs feature decomposition on the full-space permeability tensor field of the structural zone, extracts the principal direction and principal value, identifies high water-conductivity channels along the fault strike, and identifies water-retaining barriers perpendicular to the strike, characterizing the three-dimensional anisotropic water-conductivity pattern. Then, it generates a water conductivity anisotropy distribution map and registers it with the mining stress field to determine the permeability background conditions. This process improves the evaluation of structural zone water conductivity from a scalar fuzzy description to a precise tensor characterization. The spatial distribution of water-conducting channels and water-retaining barriers is objectively located, and the anisotropy identification is extended from the local borehole scale to the overall structural zone scale, improving the accuracy and reliability of the characterization of the structural zone's water control effect.

[0047] Example 6, as Figure 1 In step S3, the specific steps of stress coupling are as follows: S3.1 Mining-induced stress analysis: Establish a numerical model for coalfield mining, input coal seam occurrence conditions, mining methods and overlying strata mechanical parameters, calculate the redistribution of the three-dimensional stress field caused by mining, extract the stress tensor of the structural zone, analyze the shear stress, normal stress and shear stress ratio of the fault zone, determine the activation state of different parts of the structural zone according to the fault gouge shear slip criterion, and extract the critical shear stress ratio threshold. S3.2 Activation Region Mapping: Map the mining stress analysis results to the permeability tensor field established in step S2, mark the spatial location and geometry of potential activation regions, record the ratio of shear stress to normal stress and slip displacement in the activation regions, and construct the coupled response field of mining stress and structural zone as the driving data for permeability tensor field correction and parameter inversion in step S4 dynamic update identification.

[0048] In this embodiment: Step S3.1 establishes a numerical model for coalfield mining, inputting coal seam occurrence conditions, mining methods, and overlying strata mechanical parameters, and calculates the redistribution of the three-dimensional stress field caused by mining. The stress tensor of the structural zone is extracted, and the shear stress, normal stress, and shear stress ratio of the fault zone are analyzed. Based on the fault gouge shear slip criterion, the activation state of different parts of the structural zone is determined, and the critical shear stress ratio threshold is extracted. This process completes the task of quantitatively correlating mining stress with the mechanical response of the structural zone, overcoming the limitation of existing technologies lacking a quantitative relationship between mining stress and fault gouge permeability evolution, achieving the goal of advanced prediction of the activation state of the structural zone, and providing mechanically driven boundary conditions for the dynamic update identification in step S4.

[0049] Step S3.2 maps the mining-induced stress analysis results to the permeability tensor field established in step S2, marking the spatial location and geometry of the potential activation region, and recording the ratio of shear stress to normal stress and the amount of slip displacement in the activation region. A coupled response field between mining-induced stress and the structural zone is constructed as the driving data for the correction of the permeability tensor field and parameter inversion in the dynamic update identification step S4. This process completes the tasks of spatial positioning of the activation region and extraction of mechanical parameters, solving the technical problem that existing technologies cannot spatially correlate mining-induced stress response with the permeability tensor field, and achieving the goal of providing driving data for the dynamic correction of the permeability tensor field.

[0050] Steps S3.1 and S3.2 together construct the mining-induced stress coupling in step S3. Existing technologies lack quantitative models of the evolution of mining-induced stress and fault gouge permeability, making it impossible to predict under what mining conditions a previously water-isolated fault will transform into a water-conducting channel. Step S3 first establishes a numerical model of coalfield mining, calculates the redistribution of the three-dimensional stress field caused by mining, extracts the stress tensor of the structural zone and determines its activation state, and extracts the critical shear stress ratio threshold. Then, the analytical results of the mining-induced stress are mapped to the permeability tensor field, marking the spatial location and geometry of the activated area, recording the ratio of shear stress to normal stress and the amount of slip displacement, and constructing a coupled response field between mining-induced stress and the structural zone. This process establishes a quantitative correlation between mining-induced stress and the mechanical response of the structural zone, allowing for advanced prediction of the activation state of different parts of the structural zone, and precise location of the spatial location and mechanical parameters of the activated area. This provides reliable mechanical driving data for subsequent dynamic correction of the permeability tensor field and parameter inversion. Thus, aquifer identification is advanced from a static geological model to a mining-induced stress coupling framework, improving the accuracy and reliability of structural zone activation prediction.

[0051] Example 7, as Figure 5 In step S4, the specific steps for dynamically updating the identification are as follows: S4.1 Tensor Dynamic Reconstruction: The activated region in step S3 is used as a dynamic boundary condition to drive the reconstruction and update of the permeability tensor field in step S2. The permeability tensor of the fault gouge in the activated region is corrected according to the relationship between shear slip and permeability coefficient increment. The coupled equation of mining and seepage is established to solve the flow field evolution. S4.2 Multi-source data assimilation: Water level monitoring and microseismic event data are injected into the mining and seepage coupling model through data assimilation algorithm. The aquifer boundary and seepage parameters are jointly inverted and corrected, and the dynamically updated aquifer feature identification results are output as the benchmark data for verification and optimization in step S5.

[0052] In this embodiment: Step S4.1 uses the activated region output in step S3 as a dynamic boundary condition to drive the reconstruction and update of the permeability tensor field established in step S2. Based on the relationship between shear slip and permeability coefficient increment, the permeability tensor of the fault gouge in the activated region is corrected, and the coupled control equations of mining and seepage are established to solve for the evolution of the aquifer flow field. This process completes the task of dynamic correction of the permeability tensor in the activated region and calculation of the flow field evolution, overcoming the limitation of existing static models that cannot reflect the abrupt changes in permeability parameters under mining conditions. It achieves the goal of real-time evolution of the aquifer flow field with the mining process, providing a dynamic parameter field basis for the data assimilation and inversion in step S4.2.

[0053] Step S4.2 uses a data assimilation algorithm to inject water level monitoring data and microseismic event data into the mining-induced seepage coupling model, performing joint inversion correction on aquifer boundary morphology and permeability parameters, and outputting dynamically updated aquifer feature identification results. This process completes the task of fusing and correcting multi-source monitoring data with the coupling model, solving the technical problem that existing technologies cannot effectively correct static identification models with real-time monitoring data, achieving the goal of dynamic synchronous correction of aquifer boundaries and parameters, and providing benchmark data for verification and optimization in step S5.

[0054] Steps S4.1 and S4.2 together construct the dynamic update identification in step S4. Existing technologies generate static aquifer models based on exploration data from a specific time section. The development of overburden fracture zones and the formation of floor failure zones caused by coal seam mining cannot be reflected in real time, leading to a severe disconnect between the identification results and actual mining activities. Step S4 first uses the activated region as a dynamic boundary condition to drive the reconstruction and update of the permeability tensor field. Based on the relationship between shear slip and permeability coefficient increments, the permeability tensor of the fault gouge in the activated region is corrected. A coupled equation for mining and seepage is established to solve the flow field evolution. Then, water level monitoring and microseismic event data are injected into the coupled model through a data assimilation algorithm, jointly inverting and correcting the aquifer boundary and permeability parameters. This process forms a two-level progressive optimization mechanism of mechanical response-driven parameter updates and real-time data assimilation correction, enabling the aquifer identification results to evolve in real time with the mining process. This elevates aquifer feature identification from static lag analysis to dynamic synchronous updates, improving the timeliness and accuracy of water inrush hazard assessment.

[0055] Example 8, as Figure 6 In step S4.1, the specific steps for tensor dynamic reconstruction are as follows: S4.11 Activation Boundary Drive: The spatial coordinates and geometric shape of the activation region output in step S3 are used as dynamic boundary conditions to drive the reconstruction and update of the permeability tensor field established in step S2, determine the boundary node coordinates of the activation region and the initial value of the permeability tensor, and establish the mapping relationship between the dynamic boundary conditions and the static tensor field. S4.12, Fault gouge tensor correction: Based on the relationship model between shear slip and permeability coefficient increment in the activated region, the original low permeability tensor of the fault gouge in the activated region is corrected and a high permeability tensor is assigned. The coupled control equations of mining and seepage are established, and the evolution of the aquifer flow field is solved using the corrected tensor field as the parameter field. The pressure distribution and velocity vector field of the flow field are output.

[0056] In this embodiment: Step S4.11 uses the spatial coordinates and geometric shape of the activated region determined in step S3 as dynamic boundary conditions to drive the reconstruction and update of the permeability tensor field established in step S2. This step determines the coordinates of the boundary nodes of the activated region and the initial value of the permeability tensor, establishes the mapping relationship between the dynamic boundary conditions and the static tensor field, and transforms the mechanical response of the tectonic zone caused by mining into a spatial constraint for permeability parameter updates. This completes the task of transferring the boundary conditions of the activated region to the permeability tensor field, achieving the goal of driving the static tensor field to transform into a dynamic model.

[0057] Step S4.12, based on the relationship model between shear slip and permeability coefficient increment in the activated region, corrects the original low-permeability tensor of the fault gouge in the activated region and assigns a high-permeability tensor. Using the corrected tensor field as the parameter field, it establishes the coupled control equations for mining and seepage, solves the evolution of the aquifer flow field, and outputs the pressure distribution and velocity vector field of the flow field. This process completes the task of quantitatively correcting the abrupt change in permeability parameters in the activated region and calculating the flow field evolution, overcoming the limitation of existing technologies that cannot reflect the dramatic increase in permeability of fault gouge under mining conditions, and achieving the goal of real-time evolution of the aquifer flow field during the mining process.

[0058] Steps S4.11 and S4.12 together construct the tensor dynamic reconstruction of step S4.1. Existing technologies establish aquifer models based on static exploration data, which cannot reflect the abrupt changes in permeability parameters caused by fault activation under mining conditions. This application first uses the spatial coordinates and geometric shape of the activated region as dynamic boundary conditions to drive the reconstruction and update of the permeability tensor field and establish a mapping relationship. Then, based on the relationship between shear slip and permeability coefficient increment, the fault gouge permeability tensor is corrected, and a coupled equation of mining and seepage is established to solve the flow field evolution. This process establishes a dynamic correlation between mining stress response and permeability parameter updates, quantitatively characterizes the transformation of fault gouge from low to high permeability, and enables the aquifer flow field evolution to be calculated in real time during the mining process. This upgrades the static identification model to a dynamic update model, improving the synchronization between aquifer feature identification results and actual mining.

[0059] Example 9, as Figure 7 In step S4.2, the specific steps for multi-source data assimilation are as follows: S4.21 Monitoring Data Injection: Water level monitoring data and microseismic event data are used as observation datasets. Observation operators are constructed to establish the mapping relationship between monitoring data and model state variables. The observation dataset is integrated into the mining-induced and seepage coupling model through a data assimilation algorithm. The model state estimates are iteratively updated to reduce the deviation between model predictions and measured data. S4.22 Boundary Parameter Inversion: Based on the updated model state estimate, the boundary morphology and permeability tensor parameters of the aquifer are jointly inverted and corrected to generate dynamically updated aquifer feature identification results. These results include the spatial distribution of the aquifer, the location of the water-conducting channels, and the distribution of permeability parameters, and serve as the benchmark data for the closed-loop verification and optimization step S5.

[0060] In this embodiment: Step S4.21 uses water level monitoring data and microseismic event data as the observation dataset, constructs an observation operator to establish a mapping relationship between the monitoring data and the model state variables, and integrates the observation dataset into the mining-induced and seepage-induced coupled model through a data assimilation algorithm. The model state estimate is iteratively updated to reduce the deviation between model predictions and measured data. This process completes the task of fusing and correcting multi-source real-time monitoring data with the coupled model, overcoming the limitation of existing technologies where real-time monitoring data cannot effectively correct static identification models. It achieves the goal of reducing model prediction uncertainty and providing updated state estimates for subsequent inversion correction.

[0061] Step S4.22, based on the updated model state estimates, performs joint inversion correction on the aquifer boundary morphology and permeability tensor parameters, generating dynamically updated aquifer feature identification results that include the spatial distribution of the aquifer, the location of water-conducting channels, and the distribution of permeability parameters. This process completes the task of dynamically and synchronously correcting the aquifer boundary and parameters, solving the technical problem of lagging identification results due to the disconnect between existing monitoring data and model state. It achieves the goal of outputting dynamically updated aquifer feature identification results and provides benchmark data for closed-loop verification and optimization in step S5.

[0062] Steps S4.21 and S4.22 together construct step S4.2, multi-source data assimilation. Existing technologies establish aquifer models based on static exploration data, with real-time monitoring data only used for post-event verification and unable to be integrated into the model for dynamic correction, resulting in a severe disconnect between the identification results and actual mining activities. This application first uses water level monitoring and microseismic event data as the observation dataset, constructs observation operators to establish mapping relationships, integrates the data assimilation algorithm into the mining and seepage coupling model, iteratively updates the model state estimates to reduce prediction bias, and then performs joint inversion correction on the aquifer boundary morphology and permeability tensor parameters based on the updated state estimates, generating dynamically updated aquifer feature identification results. This process transforms real-time monitoring data from passive verification to actively driving model updates, allowing aquifer boundaries and permeability parameters to be continuously corrected during mining activities, elevating aquifer feature identification from static lag analysis to dynamic synchronous updates, and improving the consistency between the identification results and actual mining activities.

[0063] Example 10, as follows Figure 1 In step S5, the specific steps for closed-loop verification optimization are as follows: S5.1 Multi-scale verification: The dynamic aquifer feature identification results of step S4 are compared and verified with the actual water inflow points and water quality monitoring data in the well at multiple scales. The consistency index between the identification results and the actual water inflow points in three dimensions—spatial location, water inflow volume, and water chemistry type—is calculated to quantify the spatial distribution characteristics and magnitude of the identification deviation. S5.2 Deviation Source Tracing and Optimization: Based on the verification deviation, trace back to the structural model in step S1, the tensor parameters in step S2, or the stress calculation results in step S3 to locate the source of the deviation and trigger the self-optimization of the corresponding step model. Write the optimized parameters and structure into the knowledge base and feed them back to step S1 for the next round of identification, forming a closed-loop self-evolutionary system.

[0064] In this embodiment: Step S5.1 compares and verifies the dynamic aquifer feature identification results output in step S4 with the actual downhole water inflow points and water quality monitoring data at multiple scales. The degree of agreement between the identification results and the actual water inflow points is calculated in three dimensions: spatial location, water inflow magnitude, and hydrochemical type. This quantifies the spatial distribution characteristics and magnitude of the identification deviation. This process completes the task of quantitatively comparing the identification results with actual engineering data, overcoming the limitation of existing technical models lacking verification with measured data. It achieves the objective evaluation of identification accuracy and extraction of deviation information, providing a data foundation for subsequent deviation tracing.

[0065] Step S5.2 traces back to the structural model of step S1, the tensor parameters of step S2, or the stress calculation results of step S3 based on the verification deviation. The source of the deviation is located, and the corresponding step's model self-optimization is triggered. The optimized parameters and structure are written into the knowledge base and fed back to step S1 for the next round of recognition. This process completes the tasks of deviation source location and model adaptive correction, solving the technical problem of error accumulation and propagation in the existing unidirectional linear process, which cannot be corrected. It achieves the goal of forming a closed-loop self-evolving system, enabling the recognition accuracy to continuously improve with application frequency.

[0066] Steps S5.1 and S5.2 together constitute the closed-loop verification and optimization process of step S5. Existing technologies employ a unidirectional linear process, where data acquisition, fusion, identification, and evaluation are independent of each other. Errors from previous steps are propagated and cannot be corrected, making it difficult to guarantee the reliability of the identification results. Step S5 first performs multi-scale comparison and verification with actual downhole water inflow points and water quality monitoring data, calculating the consistency indices of spatial location, water inflow magnitude, and water chemistry type to quantify the identification deviation. Then, based on the deviation, it traces back to the structural model, tensor parameters, or stress calculation results to locate the source of the deviation and trigger model self-optimization. The optimized parameters and structure are written into the knowledge base and fed back to step S1. This process establishes a quantitative comparison relationship between the identification results and engineering reality, accurately locating and correcting the source of the deviation, transforming the unidirectional linear process into a closed-loop self-evolving system. As a result, the identification accuracy continuously improves with the frequency of application, enhancing the reliability and stability of aquifer feature identification.

[0067] The above description of the disclosed embodiments enables those skilled in the art to make or use the 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 invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for identifying the characteristics of coalfield aquifers, characterized in that, The specific steps are as follows: S1. Multi-source fusion modeling: Collect multi-source data, establish a spatial framework, identify fault zone boundaries and lithological zoning, and generate a three-dimensional structural model; In step S1, the specific steps of multi-source fusion modeling are as follows: S1.1 Multi-source data spatial registration: Collect three-dimensional seismic, transient electromagnetic and hydrological borehole data in the coalfield area, establish a unified spatial framework through time-depth conversion and coordinate registration, identify fault zone boundaries based on seismic wave impedance differences, delineate the influence range of tectonic zones in combination with transient electromagnetic low-resistivity anomalies, and mark the lithological zoning interfaces within the tectonic zones based on borehole cores. S1.2 Generation of three-dimensional heterogeneous structure: The registered multi-source data is input into the geostatistical stochastic simulation framework. The spatial correlation structure is established through variogram analysis under the constraint of lithological zoning, generating a heterogeneous structure model of tectonic zone. The three-dimensional distribution model of fault core, fracture zone and fracture influence zone is output and used as the input data for step S2. S2. Permeability Tensor Construction: Based on the three-dimensional structural model, permeability coefficient tensors are assigned to lithological zoning to establish a permeability tensor field, identify water-conducting channels and water-impermeable barriers, and output anisotropic distribution maps. In step S2, the specific steps for constructing the permeation tensor are as follows: S2.1 Zonal Tensor Assignment: Based on the three-dimensional structural model output in step S1, the fault core, fractured rock zone and fracture influence zone are assigned corresponding permeability tensors according to the lithological zoning characteristics of the tectonic zone. The principal direction of the permeability tensor of the fracture influence zone is calculated based on the fracture orientation data. The fracture network equivalence method is used to transform the discrete fracture distribution into the permeability tensor of the equivalent continuous medium. S2.2 Anisotropic Field Identification: Based on the zonal assignment results, a full-space permeability tensor field of the structural zone is established. Through the analysis of the principal directions and principal values ​​of the tensor, high water-conducting channels along the fault strike and water-blocking barriers perpendicular to the strike are identified. The anisotropic distribution map of water conductivity of the structural zone is output and used as input data for the stress coupling calculation of mining in step S3. S3. Mining-induced stress coupling: Establish a numerical model for coalfield mining, input coal seam occurrence conditions and overlying mechanical parameters, calculate the redistribution of the three-dimensional stress field caused by mining, extract the stress tensor of the structural zone, determine the activation risk, and output the coupled response field of mining-induced stress and structural zone. In step S3, the specific steps of stress coupling are as follows: S3.1 Mining-induced stress analysis: Establish a numerical model for coalfield mining, input coal seam occurrence conditions, mining methods and overlying strata mechanical parameters, calculate the redistribution of the three-dimensional stress field caused by mining, extract the stress tensor of the structural zone, analyze the shear stress, normal stress and shear stress ratio of the fault zone, determine the activation state of different parts of the structural zone according to the fault gouge shear slip criterion, and extract the critical shear stress ratio threshold. S3.2 Activation Region Mapping: Map the mining stress analysis results to the permeability tensor field established in step S2, mark the spatial location and geometry of potential activation regions, record the ratio of shear stress to normal stress and slip displacement in activation regions, and construct a coupled response field of mining stress and structural zone as driving data for permeability tensor field correction and parameter inversion in step S4 dynamic update identification. S4. Dynamic Update Identification: The activated region is used as a dynamic boundary condition to drive the tensor field update and correct the permeability tensor, establish the mining-driven seepage coupling equation, inject monitoring data to invert aquifer parameters, and output dynamic identification results. In step S4, the specific steps for dynamically updating the identification are as follows: S4.1 Tensor Dynamic Reconstruction: The activated region in step S3 is used as a dynamic boundary condition to drive the reconstruction and update of the permeability tensor field in step S2. The permeability tensor of the fault gouge in the activated region is corrected according to the relationship between shear slip and permeability coefficient increment. The coupled equation of mining and seepage is established to solve the flow field evolution. S4.2 Multi-source data assimilation: Water level monitoring and microseismic event data are injected into the mining and seepage coupling model through data assimilation algorithm, and the aquifer boundary and seepage parameters are jointly inverted and corrected. The dynamically updated aquifer feature identification results are output as the benchmark data for verification and optimization in step S5. S5. Closed-loop verification and optimization: The identification results are compared with the water inflow point and water quality data to calculate the consistency index. Based on the deviation, the model self-optimization is triggered by reverse tracing and the optimization parameters are fed back to form a closed-loop self-evolution system. In step S5, the specific steps for closed-loop verification optimization are as follows: S5.1 Multi-scale verification: The dynamic aquifer feature identification results of step S4 are compared and verified with the actual water inflow points and water quality monitoring data in the well at multiple scales. The consistency index between the identification results and the actual water inflow points in three dimensions—spatial location, water inflow volume, and water chemistry type—is calculated to quantify the spatial distribution characteristics and magnitude of the identification deviation. S5.2 Deviation Source Tracing and Optimization: Based on the verification deviation, trace back to the structural model in step S1, the tensor parameters in step S2, or the stress calculation results in step S3 to locate the source of the deviation and trigger the self-optimization of the corresponding step model. Write the optimized parameters and structure into the knowledge base and feed them back to step S1 for the next round of identification, forming a closed-loop self-evolutionary system.

2. The method for identifying coalfield aquifer features according to claim 1, characterized in that, In step S2.1, the specific steps for assigning values ​​to the segmented tensor are as follows: S2.11 Initialization of Lithological Zoning Tensors: Based on the three-dimensional structural model in step S1, permeability coefficient tensors of fault cores and fractured rock zones are initialized according to the lithological zoning characteristics of the tectonic zone, and permeability tensor fields of the tectonic zone matrix are constructed. The principal value of the fault core tensor corresponds to low permeability characteristics, and the principal value of the fractured rock zone tensor corresponds to medium permeability characteristics. S2.12, Superposition of equivalent fracture tensors: Extract fracture orientation data of the fracture influence zone, calculate the normal and tangential permeability coefficients of the fracture surface, and use the fracture network equivalence method to transform the discrete fracture distribution into an equivalent continuous medium permeability tensor, which is superimposed on the matrix skeleton tensor field to form a heterogeneous anisotropic permeability tensor field of the entire tectonic zone.

3. The method for identifying coalfield aquifer features according to claim 1, characterized in that, In step S2.2, the specific steps for anisotropic field identification are as follows: S2.21 Tensor Feature Analysis: Based on the full-space permeability tensor field of the tectonic zone established in step S2.1, feature decomposition is performed on the tensors of each spatial node to extract the principal direction and principal value. High water-conducting channels with dominant fracture connectivity are identified along the fault strike, and water-impermeable barriers formed by fault gouge are identified perpendicular to the fault strike, thus characterizing the three-dimensional anisotropic water-conducting pattern of the tectonic zone. S2.22 Anisotropic Mapping: Based on the tensor feature analysis results, generate anisotropic distribution map of water conductivity in the structural zone, identify the spatial location of water-conducting channels, the distribution range of water barriers, and the permeability tensor parameters of nodes, and register the distribution map with the mining stress field in spatial coordinates to determine the background conditions of permeability in the structural zone and use it as input data for the mining stress coupling calculation in step S3.

4. The method for identifying coalfield aquifer features according to claim 1, characterized in that, In step S4.1, the specific steps for tensor dynamic reconstruction are as follows: S4.11 Activation Boundary Drive: The spatial coordinates and geometric shape of the activation region output in step S3 are used as dynamic boundary conditions to drive the reconstruction and update of the permeability tensor field established in step S2, determine the boundary node coordinates of the activation region and the initial value of the permeability tensor, and establish the mapping relationship between the dynamic boundary conditions and the static tensor field. S4.12, Fault gouge tensor correction: Based on the relationship model between shear slip and permeability coefficient increment in the activated region, the original low permeability tensor of the fault gouge in the activated region is corrected and a high permeability tensor is assigned. The coupled control equations of mining and seepage are established, and the evolution of the aquifer flow field is solved using the corrected tensor field as the parameter field. The pressure distribution and velocity vector field of the flow field are output.

5. The method for identifying coalfield aquifer features according to claim 1, characterized in that, In step S4.2, the specific steps of multi-source data assimilation are as follows: S4.21 Monitoring Data Injection: Water level monitoring data and microseismic event data are used as observation datasets. Observation operators are constructed to establish the mapping relationship between monitoring data and model state variables. The observation dataset is integrated into the mining-induced and seepage coupling model through a data assimilation algorithm. The model state estimates are iteratively updated to reduce the deviation between model predictions and measured data. S4.22 Boundary Parameter Inversion: Based on the updated model state estimate, the boundary morphology and permeability tensor parameters of the aquifer are jointly inverted and corrected to generate dynamically updated aquifer feature identification results. These results include the spatial distribution of the aquifer, the location of the water-conducting channels, and the distribution of permeability parameters, and serve as the benchmark data for the closed-loop verification and optimization step S5.

Images