Dynamic reconstruction method of gas occurrence in complex coal seam group based on space-time evolution field theory

Abstract

The present application relates to the technical field of coal mine safety engineering and geological disaster prevention, and discloses a complex coal seam group gas occurrence dynamic reconstruction method based on space-time evolution theory, comprising: S1: constructing a multi-scale geological structure evolution basic model; S2: establishing a double medium gas occurrence parameter space-time evolution equation; S3: performing multi-source data dynamic correction and multi-scale Bayesian inversion reconstruction; wherein, step S1 provides a permeability distribution model and geological constraints for step S2, step S2 provides a forward core framework for step S3, and step S3 corrects the evolution equation parameters of step S2 through real-time data in reverse, forming a closed loop linkage. The present application realizes high-precision, real-time dynamic reconstruction of the gas occurrence state of complex structure coal seams, effectively improving the precision decision-making ability of mine gas disaster prevention.

Description

Technical Field

[0001] This invention relates to the fields of coal mine safety engineering and geological disaster prevention technology, specifically to a dynamic reconstruction method for gas occurrence in complex coal seam groups based on spatiotemporal evolution field theory. Background Technology

[0002] Accurate perception of the occurrence state of coal seam gas is a core prerequisite for the prevention and control of coal mine gas disasters. Complex coal seam groups generally have characteristics such as well-developed geological structures, strong spatiotemporal heterogeneity of coal seam permeability, and significant coupling between mining stress field and gas seepage field, which leads to the gas occurrence exhibiting a strongly nonlinear and multi-scale dynamic change law.

[0003] Existing gas occurrence reconstruction methods mostly rely on static geological modeling and single-scale monitoring data, which have the following shortcomings: First, the data processing method is singular and does not consider the multi-scale characteristics of time-series data such as gas emission, making it difficult to effectively separate large-scale background trends from small-scale anomalous fluctuations, and it is easily affected by observation noise. Second, the inversion framework lacks targeted constraints. Traditional inversion methods mostly use single smoothing constraints or sparse constraints, which cannot take into account the spatial continuity of gas occurrence and local anomalous characteristics, resulting in limited reconstruction accuracy. Third, it is not adaptable enough to complex structures and the dynamic impact of mining, making it difficult to achieve real-time dynamic updates of gas occurrence status, resulting in deviations between the reconstruction results and the actual situation on site, and failing to meet the needs of precise prevention and control.

[0004] Although some methods attempt to improve stability by introducing Bayesian inversion theory, they still cannot solve the problems of accuracy and dynamism in reconstructing gas occurrence in complex coal seam groups without combining multi-scale data decomposition technology and designing differentiated constraint mechanisms for the spatial distribution characteristics of gas occurrence. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a dynamic reconstruction method for gas occurrence in complex coal seam groups based on spatiotemporal evolution field theory. This method can achieve high-precision, real-time dynamic monitoring and reconstruction of gas occurrence status, providing technical support for gas extraction optimization, mine ventilation adjustment, and safety production decision-making. It is applicable to coal mines with complex geological structures (faults, folds, collapse columns, etc.).

[0006] To achieve the above objectives, the following technical solution is adopted: a dynamic reconstruction method for gas occurrence in complex coal seam groups based on spatiotemporal evolution field theory, comprising: S1: establishing a three-dimensional geological model of the coal seam based on geological structural data, and generating an initial permeability distribution field of the coal seam that matches the structural space by quantifying the curvature and interface complexity of the structure; S2: using the initial permeability distribution field as input, establishing a spatiotemporal evolution model describing the occurrence and migration of gas in the matrix and fractures based on dual-medium theory, wherein the spatiotemporal evolution model is coupled with a dynamic reconstruction method based on mining stress. S3: Real-time correction of dynamic damage mechanism of fracture permeability; S4: Collect real-time monitoring data of mine gas outburst, and extract its different frequency characteristic components through multi-scale signal decomposition technology; construct a Bayesian inversion framework with the spatiotemporal evolution model as the physical forward modeling constraint; wherein, the Bayesian inversion framework implements differentiated spatial constraints on the inversion parameters of the mining-affected area and the non-affected area by incorporating spatially variable prior information; by solving the framework, the gas occurrence parameter field is dynamically reconstructed and used to update the spatiotemporal evolution model to achieve closed-loop correction;

[0007] Specifically, the differentiated spatial constraints on the inversion parameters of the mining-affected and non-affected areas are achieved by constructing a spatially variable covariance prior distribution. This prior distribution is set differently based on the spatial relationship between the grid nodes and the mining face: for grid nodes within the dynamic mining-affected area, a first numerical range of prior variance is set; for grid nodes outside the affected area, a second numerical range of prior variance is set, and spatial smoothing constraints are applied to maintain regional continuity. The first numerical range is greater than the second numerical range, and the dynamic mining-affected area is dynamically determined based on the real-time advance speed of the working face.

[0008] Furthermore, in step S1, generating the initial permeability distribution field includes: quantifying the curvature and interface complexity of the structure, specifically: calculating the structural curvature σ reflecting the degree of structural bending based on seismic interpretation data, and calculating the fractal dimension D reflecting its irregularity based on the structural interface data exposed by boreholes; establishing a permeability control model K(σ,D) with the structural curvature σ and fractal dimension D as variables based on statistical regression methods; and substituting the values ​​of structural curvature σ and fractal dimension D at each spatial location into the permeability control model K(σ,D) to calculate and generate the initial permeability distribution field.

[0009] Furthermore, in step S1, when constructing the three-dimensional geological model, tetrahedral meshes are used for subdivision, and different mesh sizes are set for densely structured areas and regular areas, wherein the mesh size of densely structured areas is smaller than that of regular areas.

[0010] Furthermore, in step S2, the spatiotemporal evolution model is a dynamic model based on the dual-medium theory. Its construction includes: dividing the coal seam into a coal matrix system and a fracture system, and establishing respectively: the governing equations describing the free gas pressure diffusion in the fracture system; the governing equations describing the dynamics of adsorbed gas content in the coal matrix system; and the coupling relationship describing the gas mass exchange between the coal matrix system and the fracture system.

[0011] Furthermore, the spatiotemporal evolution model is coupled with the mining-induced damage mechanism to correct the permeability of the fracture system in real time according to the mining process. Specifically, it includes: calculating the damage variables and volumetric strain of the coal body based on the mining-induced stress; dynamically correcting the porosity of the fracture system based on the damage variables and volumetric strain; and updating the permeability of the fracture system in real time using the corrected porosity according to the fracture cubic law.

[0012] Furthermore, the spatiotemporal evolution model achieves numerical simulation by simultaneously solving the following governing equations: the gas pressure diffusion equation of the fracture system based on gas pseudo-pressure linearization; the gas exchange rate equation between the matrix system and the fracture system based on the linear driving force assumption; the content kinetic equation of adsorbed gas in the matrix system; and the permeability evolution equation of the fracture system coupled with mining-induced damage.

[0013] Furthermore, in step S3, different frequency feature components are extracted through multi-scale signal decomposition technology. Specifically, the gas emission time series data is decomposed into different frequency components that characterize large-scale background changes, meso-scale transition changes, and small-scale abnormal fluctuations based on the matching pursuit principle. The decomposed components are then weighted and fused based on the weights determined by historical data verification to form a fused data vector for inversion.

[0014] Furthermore, in step S3, the differentiated spatial constraints on the inversion parameters are specifically implemented by constructing a spatially variable covariance prior distribution. This prior distribution is set differently based on the spatial relationship between the grid nodes and the mining face: for grid nodes within the dynamic mining influence range, a first numerical range of prior variance is set; for grid nodes outside the influence range, a second numerical range of prior variance is set, and spatial smoothing constraints are applied to maintain regional continuity; wherein, the first numerical range is greater than the second numerical range, and the dynamic mining influence range is dynamically determined according to the real-time advance speed of the working face.

[0015] Furthermore, the Bayesian inversion framework constructed in step S3 has a posterior probability distribution maximization problem that is equivalent to the following optimization problem:

[0016]

[0017] in, To obtain the optimal estimate of the gas content parameters to be inverted, The forward modeling operator is constructed based on the aforementioned spatiotemporal evolution model. For the observed data vector, Fit residual terms to the weighted data; For spatial variation covariance matrix Prior constraints, is a sparse regularization term used to facilitate the characterization of local anomalies; m is the vector of parameters to be inverted; : Parameter vector to be inverted The transpose of .

[0018] Furthermore, the optimization problem is numerically solved using the alternating direction multiplier method to obtain the optimal estimate of the gas content parameter.

[0019] Compared with the prior art, the present invention achieves the following beneficial effects:

[0020] 1. High accuracy of multi-scale data decomposition: By combining the MPSTFT algorithm with the Ricker wavelet overcomplete dictionary, it can adaptively match the multi-scale features of gas emission time series data, effectively separate different frequency components, and eliminate noise interference. Compared with traditional single-scale data processing methods, the data utilization and purity are improved by more than 20%, providing a reliable data foundation for inversion.

[0021] 2. The inversion framework is highly scientific: The Bayesian multi-scale inversion, which integrates sparse constraints and smooth background constraints, not only suppresses the fluctuation of invalid parameters and highlights local gas anomalies through L1 regularization, but also ensures the overall spatial continuity through spatial variable covariance prior. This solves the contradiction of traditional inversion, which is "either smooth over-smoothing or abnormal distortion", and the relative reconstruction error is controlled within 15%.

[0022] 3. Excellent dynamic adaptability: By dynamically calculating the radius of influence of mining and constructing the spatial variable covariance matrix, the inversion constraints can be adjusted in real time. It can accurately respond to the impact of mining progress and structural changes on gas occurrence. Compared with the static reconstruction method, the dynamic update delay is shortened to the hour level, which meets the needs of real-time decision-making on site.

[0023] 4. High engineering practicality: It integrates multi-source geological and monitoring data, without the need for additional complex monitoring equipment. Data acquisition can be achieved based on the existing seismic exploration, drilling, and extraction ventilation monitoring systems in the mine. The phased implementation strategy lowers the threshold for engineering application and is suitable for various complex coal seam groups, with significant promotional value.

[0024] In summary, this invention constructs a complete technical chain from geological modeling to evolution equations, multi-scale inversion, and dynamic correction, deeply integrating multi-scale data processing with Bayesian inversion to form a standardized process from data preprocessing to result output, thereby improving the stability and repeatability of the method.

[0025] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of the present invention, nor is it intended to restrict the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0026] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the invention. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein:

[0027] Figure 1 This is a flowchart illustrating the dynamic reconstruction method for gas occurrence in complex coal seam groups based on spatiotemporal evolution field theory, according to an embodiment of the present invention.

[0028] Figure 2 This is a schematic diagram of three-dimensional geological visualization according to an embodiment of the present invention;

[0029] Figure 3 This is a schematic diagram of the multi-scale signal decomposition and data fusion process in an embodiment of the present invention;

[0030] Figure 4 This is a schematic diagram illustrating the construction and solution of the Bayesian inversion framework in an embodiment of the present invention;

[0031] Figure 5 This is a schematic diagram illustrating the signal reconstruction performance of the algorithm in an embodiment of the present invention. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some 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.

[0033] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0034] The dynamic reconstruction method for gas occurrence in complex coal seam groups based on spatiotemporal evolution field theory provided in this invention takes spatiotemporal evolution field theory as its core, integrates multi-source geological and engineering data, and achieves high-precision dynamic reconstruction of gas occurrence state in complex coal seam groups through a progressive logic of basic model construction, evolution law modeling, and real-time data correction.

[0035] Figure 1 This is a flowchart illustrating the dynamic reconstruction method for gas occurrence in complex coal seam groups based on spatiotemporal evolution field theory, according to an embodiment of the present invention. Figure 1 As shown, a dynamic reconstruction method for gas occurrence in complex coal seam groups based on spatiotemporal evolution field theory is presented. Step S1 constructs a multi-scale geological structural evolution model, providing core parameters (such as permeability) and geological constraints for the gas spatiotemporal evolution equation in step S2. The dual-medium evolution equation established in step S2 forms the core framework for the forward modeling in step S3's dynamic inversion. Step S3 uses multi-source real-time data and Bayesian inversion technology to reverse-correct the evolution equation parameters from step S2, achieving dynamic model optimization. These three steps form a closed-loop linkage relationship of parameter constraints, equation construction, and real-time correction. The method specifically includes the following steps:

[0036] S1: Based on geological structural data, a three-dimensional geological model of the coal seam is established. By quantifying the curvature and interface complexity of the structure, an initial permeability distribution field of the coal seam that matches the structural space is generated.

[0037] Step S1 is used to construct a basic model of multi-scale geological structural evolution. As the geological foundation of the entire method, Step S1 provides accurate spatial parameter distributions for subsequent gas flow simulation by quantifying the correlation between structural features and permeability.

[0038] Specifically, step S1, based on the theory of tectonic evolution and the concept of multi-scale geological modeling, integrates macroscopic regional tectonic data and microscopic mine measured data. By quantifying the correlation between tectonic complexity and coal seam permeability, a gridded permeability distribution model is established to ensure the spatial matching of geological structural characteristics and seepage parameters, providing core parameter constraints for the gas seepage equation. Specifically, quantifying the curvature and interface complexity of the structure involves: calculating the tectonic curvature σ, reflecting the degree of structural bending, based on seismic interpretation data, and calculating the fractal dimension D, reflecting its irregularity, based on borehole-exposed tectonic interface data; establishing a permeability control model K(σ,D) with the tectonic curvature σ and fractal dimension D as variables using statistical regression methods; and substituting the tectonic curvature σ and fractal dimension D values ​​at various spatial locations into the permeability control model K(σ,D) to calculate and generate the initial permeability distribution field. The specific implementation is as follows:

[0039] S11: Data Preprocessing

[0040] We collected geological tectonic evolution history data for the entire province / mining area (tectonic movement periods, stress directions, and evolution durations), 3D seismic data from the mine (full-band seismic records), and borehole data (coordinates, coal seam thickness, lithology, measured permeability values, and tectonic interface sampling points). We used a matching pursuit algorithm to denoise the seismic data and kriging interpolation to complete missing values ​​in the borehole data. We unified the spatial coordinate system of all data (using the mine's independent coordinate system).

[0041] S12: Initial 3D Geological Model Construction

[0042] Using seismic data stratigraphic interpretation results and borehole longitudinal constraints, an initial 3D geological model incorporating faults, folds, and collapse columns was constructed using tetrahedral mesh generation technology. Different mesh sizes were set for densely structured areas and conventional areas, with the mesh size in densely structured areas being smaller than that in conventional areas. In a preferred embodiment, the mesh size in densely structured areas was set to 3-5m, and in conventional areas to 10-20m, ensuring a balance between structural detail and computational efficiency. Specifically, the 3-5m mesh size in densely structured areas balances the accuracy of structural detail representation with model computational efficiency; the 10-20m mesh size in conventional areas is used to improve computational efficiency in areas with simple structures. Figure 2 The image shown is a schematic diagram of three-dimensional geological visualization according to an embodiment of the present invention.

[0043] S13: Quantifying Construction Curvature and Interface Complexity

[0044] (1) Construction curvature Calculation: For each grid node, through its surroundings Seismic interpretation horizon data within the neighborhood were used to calculate tectonic curvature using the second derivative method. .

[0045] Where σ: structural curvature, in units of 1 / m, used to quantify the degree of curvature of geological structures; z: stratigraphic depth, i.e., the vertical depth coordinate of geological strata in three-dimensional space; x: horizontal coordinate, one of the horizontal coordinates in the three-dimensional geological model; y: horizontal coordinate, the other of the horizontal coordinates in the three-dimensional geological model.

[0046] (2) Fractal dimension Calculation: Based on the sampling points of the structural interface revealed by the borehole, the fractal dimension is fitted using the box-counting method: .

[0047] Where D: fractal dimension, ranging from 1.2 to 1.8, is used to characterize the complexity of the constructed interface; ε: box side length, the side length of the virtual box used to cover the constructed interface in the box dimension method; : The minimum number of boxes required to completely cover the target construction interface, that is, the minimum number of boxes required to completely cover the target construction interface when the side length of the box is ε.

[0048] S14: Establishment of Permeability Control Model

[0049] Based on statistical regression analysis of borehole measured permeability and structural parameters, and calibrated with laboratory core experimental data, a permeability control function is established:

[0050]

[0051] The function was obtained by fitting using the least squares method, and the goodness of fit was... .

[0052] in, : Permeability control function, used to calculate the permeability distribution of coal seam grids, and the output value is the coal seam permeability of the corresponding grid node; : The basic value of the original coal seam permeability, in mD, is taken as the average value of the measured permeability data from the borehole, and is used as the benchmark value for permeability calculation; R²: goodness of fit, used to evaluate the fitting effect of the permeability control function, and is required to be ≥0.85 to ensure the function's fit to the measured data.

[0053] S2: Using the initial permeability distribution field as input, a spatiotemporal evolution model describing the occurrence and migration of gas in the matrix and fractures is established based on the dual-medium theory. The spatiotemporal evolution model is coupled with a dynamic damage mechanism driven by mining stress to correct fracture permeability in real time.

[0054] Step S2 is used to establish the spatiotemporal evolution equation of the dual-medium gas occurrence parameters.

[0055] Step S2 uses the initial permeability distribution field output from step S1 as the core parameter input to construct a dynamic evolution mathematical model of gas occurrence, which is the core bridge connecting the geological basis and dynamic reconstruction.

[0056] Step S2 is based on the seepage theory and adsorption kinetics principle of dual media (coal matrix-fracture system), which divides the dynamic equilibrium process of free gas in the fracture system and adsorbed gas in the matrix. The fracture cubic law and volumetric strain coupling model are used to describe the influence of mining on permeability. The spatiotemporal evolution control equation and spatiotemporal evolution model of gas pressure and content are constructed to accurately characterize the gas desorption-seepage coupling characteristics of the tectonic coal area.

[0057] S21: Basic parameter determination

[0058] Key parameters, including adsorption and desorption parameters (Langmuir pressure), were determined through laboratory experiments. Langmuir volume Adsorption time constant ), physical parameters of coal (matrix porosity) Initial fracture porosity Gas dynamic viscosity Gas compressibility factor Z), mechanical parameters (elastic modulus E, Poisson's ratio) Volumetric strain coefficient Damage coupling coefficient Stress sensitivity coefficient ).

[0059] Langmuir pressure, measured in MPa, with a range of values. , is the core parameter describing the adsorption characteristics of coal seam gas, reflecting the relationship between pressure and adsorption amount during the adsorption process; Langmuir volume, in units of value range This characterizes the maximum adsorption capacity of coal seams for methane. Adsorption time constant, in hours, with a range of values. This reflects the kinetic rate of the adsorption-desorption process of coal seam gas; Matrix porosity, value range This refers to the proportion of the volume of micropores, mesopores, and other pores within the coal matrix to the total volume of the coal body. Initial fracture porosity, value range This refers to the proportion of the volume of natural fractures in a coal seam to the total volume of the coal body when it is not affected by mining. : Gas dynamic viscosity, unit is Values ​​under certain conditions This characterizes the internal frictional resistance when methane flows in a coal seam. Gas compressibility factor, range of values It is used to correct the deviation between real gas and ideal gas, and is determined according to the coal seam pressure range; Elastic modulus, unit is GPa, range of values This characterizes the ability of coal to resist elastic deformation; Poisson's ratio, range of values It reflects the ratio of transverse strain to longitudinal strain when the coal body is deformed under stress; Volumetric strain coefficient, range of values This is used to correlate the changes in coal body volumetric strain and fracture porosity. Damage coupling coefficient, range of values The weighting of the influence of coal body damage degree on fracture porosity; Stress sensitivity coefficient, unit: value range This reflects the sensitivity of coal seam permeability to changes in effective stress.

[0060] S22: Construction of the dual-medium governing equations

[0061] The spatiotemporal evolution model in step S2 is a dynamic model based on the dual-medium theory. Its construction includes: dividing the coal seam into a coal matrix system and a fracture system, and establishing: a governing equation describing the free gas pressure diffusion within the fracture system; a governing equation describing the dynamics of adsorbed gas content within the coal matrix system; and a coupling relationship describing the gas mass exchange between the coal matrix system and the fracture system. The spatiotemporal evolution model couples with the mining-induced damage mechanism to correct the permeability of the fracture system in real time according to the mining process. Specifically, this includes: calculating the damage variables and volumetric strain of the coal body based on the mining-induced stress; dynamically correcting the porosity of the fracture system based on the damage variables and volumetric strain; and updating the permeability of the fracture system in real time using the corrected porosity according to the fracture cubic law. Furthermore, the spatiotemporal evolution model achieves numerical simulation by simultaneously solving the following governing equations: 1) Gas pressure diffusion equation of the fracture system based on gas pseudo-pressure linearization; 2) Gas exchange rate equation between the matrix system and the fracture system based on the linear driving force assumption; 3) Kinetic equation of adsorbed gas content in the matrix system; 4) Permeability evolution equation of the fracture system coupled with mining-induced damage.

[0062] (1) Pressure diffusion equation for fracture system (based on gas pseudo-pressure linearization):

[0063]

[0064]

[0065] Free gas pressure in a fracture system, measured in MPa, characterizes the pressure state of free gas in coal seam fractures; Time, measured in hours, represents the dynamic evolution of the gas occurrence state over time. Pressure transmission coefficient, used to describe the propagation rate of gas pressure in a fractured system; The Laplace operator is used to calculate the distribution gradient of fracture gas pressure in three-dimensional space. Source-sink conversion coefficient, used to convert the matrix-fracture gas mass exchange rate into source-sink terms in the pressure diffusion equation; Dynamic fracture porosity varies with mining stress and damage state, and is the result of the combined effects of initial fracture porosity, volumetric strain, and damage variables. : Matrix-fracture gas mass exchange rate, in units of Based on the linear driving force assumption, the derivation characterizes the mass of gas migrating from the matrix to the fractures per unit volume of coal per unit time. Dynamic fracture permeability, measured in mD, changes dynamically with mining stress and fracture porosity, and is a core parameter for gas flow in fractures. Overall compression ratio, in units of It comprehensively reflects the compressibility characteristics of coal and gas. It already includes gas compressibility, which conforms to the general definition of real gas seepage; Molar mass of methane, in units of Values That is, the molar mass of methane, the main component of gas; Universal gas constant, in units of Values , is a universal constant characterizing the physical properties of gases; Coal seam temperature, in Kelvin (K), taken from borehole temperature measurement data, range of values. .

[0066] (2) Matrix-fracture gas exchange rate (based on the linear driving force assumption):

[0067]

[0068] Physical meaning of matrix-fracture gas exchange rate: The mass of gas exchanged from the matrix to the fracture per unit volume of coal per unit time, with dimensions of _____. The derivation is based on Fick's diffusion law and the derivative of the Langmuir equation (the sensitivity of adsorption amount to pressure).

[0069] in, Matrix diffusion coefficient, in units of value range Characterizes the diffusion rate of gas in the micropores of the coal matrix; : Matrix particle radius, in units of value range , which is the average radius of the coal matrix particles; Density of methane gas in the fissure, in units of It is related to the pressure and temperature of the gas in the fracture and reflects the concentration of gas in the fracture. The equilibrium pressure of the adsorbed gas in the matrix is ​​expressed in MPa. It is determined by solving the Langmuir equation and the matrix gas state equation simultaneously, and represents the pressure at which the adsorbed gas in the matrix reaches equilibrium.

[0070] (3) Kinetic equation for adsorbed gas content in coal matrix system:

[0071]

[0072]

[0073] Current adsorbed methane content of the matrix, in units of This refers to the total amount of gas actually adsorbed by the coal matrix at present, which changes dynamically over time. Time, in units of This characterizes the dynamic evolution of the content of adsorbed methane in the matrix over time. Adsorption time constant, in units of value range This reflects the rate at which the adsorbed gas in the matrix relaxes to an equilibrium state. : Matrix adsorption gas equilibrium content, unit is This refers to the maximum amount of methane that the substrate can adsorb under the current equilibrium pressure. Langmuir volume, in units of value range This characterizes the maximum adsorption capacity of coal seams for methane. Coal density, in units of It characterizes the density of coal and is used to calculate the amount of gas adsorbed per unit volume of coal. Matrix porosity, value range It refers to the proportion of the volume of micropores, mesopores, and other pores inside the coal matrix to the total volume of the coal body.

[0074] (4) Mining-induced damage coupled with permeability model (based on the fracture cubic law):

[0075] Furthermore, the spatiotemporal evolution model couples a dynamic damage mechanism driven by mining stress to real-time correct fracture permeability. Specifically, this includes: calculating the volumetric strain and damage variables of the coal body based on real-time mining stress; dynamically correcting the porosity of the fracture system based on the volumetric strain and damage variables using a pre-defined relationship; and updating the permeability of the fracture system in real-time based on the corrected fracture porosity according to the fracture cubic law. The specific calculation process is as follows:

[0076]

[0077]

[0078] in, Given the current vertical stress or maximum principal stress, this model is applicable to stress relief damage scenarios (such as the goaf side). (During depressurization) Starting from 0, the damage gradually develops; for pressure-boosting scenarios (such as advanced support pressure zones), a separate damage evolution formula needs to be defined. Dynamic fracture permeability, in mD, varies with spatial coordinates ( The dynamic changes of ) and time t represent the gas permeability of coal seam fractures during mining; : Permeability control function, based on constructed curvature The basic permeability is calculated using the fractal dimension D; Dynamic fracture porosity, which varies with mining stress and damage state, is the real-time proportion of the coal seam fracture volume to the total coal volume. Initial fracture porosity, value range Porosity refers to the porosity of natural fractures in a coal seam when it is not affected by mining. Volumetric strain coefficient, range of values This is used to correlate the changes in coal body volumetric strain and fracture porosity. Volumetric strain, dimensionless, characterizes the degree of volumetric deformation of coal under mining stress; Damage coupling coefficient, range of values The weighting of the influence of coal body damage degree on fracture porosity; Damage variable, range of values This characterizes the degree of damage to the coal body under mining stress, with 0 representing no damage and 1 representing complete damage. : Mining stress at time t, in MPa, refers to the magnitude of stress acting on the coal seam at a certain moment during the mining process, specifically the vertical stress or the maximum principal stress; Damage threshold stress, in MPa, with a range of values. , refers to the critical stress value at which the coal body begins to suffer damage; Elastic modulus, unit is GPa, range of values This characterizes the ability of coal to resist elastic deformation; Poisson's ratio, range of values It reflects the ratio of transverse strain to longitudinal strain when the coal body is deformed under stress; Stress sensitivity coefficient, unit: value range This reflects the sensitivity of the degree of coal body damage to changes in stress; Horizontal coordinate, one of the horizontal coordinates in a three-dimensional geological model; : Horizontal coordinate, the second horizontal coordinate in a three-dimensional geological model; Vertical coordinates: Vertical depth coordinates in a three-dimensional geological model.

[0079] S23: Equation for calculating gas content:

[0080]

[0081] in, Total coal seam gas content, in units of , refers to spatial coordinates ( The total amount of gas contained in a unit mass of coal seam at time t is equal to the sum of the free gas content in the fractures and the gas content adsorbed by the matrix. : Free gas content in fissures, in units of This refers to the content of free-state methane in coal seam fractures; : Matrix adsorption gas content, unit is This refers to the content of adsorbed methane in the coal matrix; Dynamic fracture porosity, which varies with mining stress and damage state, is the real-time proportion of the coal seam fracture volume to the total coal volume. Molar mass of methane, in units of Values That is, the molar mass of methane, the main component of gas; Free gas pressure in a fracture system, measured in MPa, characterizes the pressure state of free gas in coal seam fractures.

[0082] S3: Collect real-time monitoring data of mine gas outbursts and extract their different frequency characteristic components through multi-scale signal decomposition technology; construct a Bayesian inversion framework using the spatiotemporal evolution model as the physical forward modeling constraint; wherein, the Bayesian inversion framework implements differentiated spatial constraints on the inversion parameters of the mining-affected area and the non-affected area by incorporating spatially variable prior information; by solving the framework, dynamically reconstruct the gas occurrence parameter field and use it to update the spatiotemporal evolution model to achieve closed-loop correction.

[0083] Step S3 is used to realize dynamic correction of multi-source data and multi-scale Bayesian inversion reconstruction.

[0084] Step S3 uses the evolution equation from step S2 as the basis for forward modeling, integrates real-time data from multiple sources to dynamically correct the model, and finally achieves high-precision reconstruction of the gas occurrence state.

[0085] Furthermore, step S3, based on the multi-scale decomposition technique of sparse representation theory and the Bayesian inversion framework, integrates multi-source real-time monitoring data. Through multi-scale data weighted fusion (weight allocation based on historical data verification) and spatial variable covariance prior construction, it achieves local high-precision dynamic reconstruction of the gas occurrence model, while ensuring the mathematical rigor and computational efficiency of the inversion process.

[0086] S31: Multi-source real-time data acquisition and preprocessing

[0087] S311: Data Acquisition: Gas Drainage Metering Data (Drainage Flow Rate) Sampling concentration Extraction pressure ), ventilation gas data (air volume) Ventilation concentration New geological information (location and attitude of minor faults) was revealed at the working face.

[0088] Extraction flow rate refers to the volumetric flow rate of gas extracted by the gas extraction system per unit time, and is one of the core data for gas extraction metering. Extraction concentration refers to the volume fraction of methane in the extracted gas, used to calculate the pure amount of extracted gas. Extraction pressure refers to the pressure inside the pipeline during the gas extraction process, reflecting the working status of the extraction system. Air volume refers to the volumetric airflow rate per unit time through the working face in a mine ventilation system. Ventilation concentration refers to the volume fraction of methane carried in the ventilation airflow and is used to calculate the amount of methane discharged.

[0089] S312: Preprocessing: Time-series smoothing was performed using the moving average method (window size 30 min); coordinate calibration was performed on newly revealed geological information, and local structural parameters were updated. ) and penetration rate The 30-minute window size for the moving average method refers to smoothing time-series data using a 30-minute time window to reduce the impact of random fluctuations.

[0090] Construction curvature, in units of Based on newly revealed geological information, it is used to quantify the degree of local tectonic curvature; Fractal dimension, range of values Based on newly revealed geological information, it is used to characterize the complexity of local tectonic interfaces; Coal seam permeability, in mD, based on updated structural parameters. Corrected local coal seam permeability.

[0091] S32: Multiscale Data Decomposition and Weighted Fusion

[0092] In step S32, different frequency feature components are extracted using multi-scale signal decomposition technology. Specifically, based on the matching pursuit principle, the gas emission time series data is decomposed into different frequency components representing large-scale background changes, mesoscale transitional changes, and small-scale abnormal fluctuations. The decomposed components are then weighted and fused based on weights determined through historical data verification to form a fused data vector d for inversion. Figure 3 The diagram illustrates the multi-scale signal decomposition and data fusion process in an embodiment of the present invention. Specifically, it includes the following steps:

[0093] S321: Analysis of gas emission time-series data based on the Matching Pursuit Short-Time Fourier Transform (MPSTFT) algorithm Multi-scale decomposition is performed, taking into account both time and frequency resolution. An overcomplete Ricker wavelet dictionary is constructed. The wavelet function is:

[0094]

[0095] Total gas emission, in units of This is equal to the sum of the extracted gas purity and the ventilation gas purity, i.e. . The Ricker wavelet overcomplete dictionary consists of multiple Ricker wavelets with different parameters, used to sparsely represent time series data of gas outbursts. Ricker wavelet functions are the basic atoms for constructing a complete dictionary, used to match different frequency components in gas emission data. : Wavelet parameter set, including time offset ,frequency and phase Three control parameters. Time offset: The time position parameter of the Ricker wavelet in the time series data, used to adjust the time alignment position of the wavelet. Frequency: The center frequency of the Ricker wavelet, used to match components of the corresponding frequency in the data. Phase: The phase parameter of the Ricker wavelet, used to optimize the matching degree between the wavelet and the data. Time: The time coordinate of time series data. The imaginary unit is used to construct complex exponential forms and characterize the phase properties of wavelets.

[0096] S322: Using a matching pursuit iterative algorithm (number of iterations) Error threshold ) decomposition yields low frequency ( Large-scale background), mid-frequency Mesoscale transition), high frequency ( (Small-scale anomaly) component.

[0097] Number of iterations The upper limit of the matching pursuit algorithm's iterations ensures a balance between algorithm efficiency and decomposition effectiveness; The error threshold of the matching pursuit algorithm; iteration stops when the decomposition residual is less than this value to ensure decomposition accuracy; the frequency range of low-frequency components. The large-scale background variation trend of gas emission; the frequency range of the mid-frequency component. The frequency range of the high-frequency component is >20Hz, which corresponds to the mesoscale transitional variation characteristics of gas emission; the frequency range of the high-frequency component is >20Hz, which corresponds to the small-scale abnormal fluctuations of gas emission.

[0098] S323: Component-weighted fusion: Weights are determined through cross-validation of historical data: First, regression analysis is performed on historical monitoring data and the gas content variation field verified by intensive borehole drilling to obtain the contribution of each scale component (regression coefficient), which is then normalized and used as the weight (low-frequency component weight). Mid-frequency component weighting High-frequency component weights A fused data vector d is constructed, which is obtained by weighting the components at different scales according to their corresponding weights. This fused data vector d is used as the input of the observation data for subsequent Bayesian inversion.

[0099] Among them, the low-frequency component weight The contribution of large-scale background information to model correction was determined through cross-validation using historical data; the mid-frequency component weights... The contribution of mesoscale transition information to model correction was determined through cross-validation using historical data; high-frequency component weights. The contribution of small-scale anomalies to model correction was determined through cross-validation using historical data.

[0100] Through step S32, this embodiment of the invention constructs a complete wavelet dictionary based on MPSTFT (Matched Pursuit Short Time Fourier Transform) and Ricker wavelet multi-scale decomposition technology, realizes multi-scale separation of gas outburst time series data, accurately extracts large-scale background components, meso-scale transition components and small-scale anomaly components, and provides high-quality data support for subsequent inversion.

[0101] S33: Construction of Bayesian Inversion Framework

[0102] like Figure 4 The diagram shown illustrates the construction and solution of the Bayesian inversion framework in an embodiment of the present invention. The construction of the Bayesian inversion framework includes the following steps:

[0103] S331: Construction of the forward operator G

[0104] The finite volume method is used to spatially discretize (using the same mesh as in step S1) and temporally differ (time step) the dual-medium governing equations in step S2. The predicted gas emission rate is obtained through numerical solution, forming a forward modeling operator. , where m is the grid parameter vector of the gas content to be inverted.

[0105] forward operator The dual-medium control equations in step S2 are constructed by discretizing them using the finite volume method, thereby mapping the parameter vector to be inverted to the predicted gas emission rate. The finite volume method is a numerical solution algorithm used to spatially discretize (with the same grid as step S1) and temporally differ the dual-medium control equations to achieve numerical simulation of the gas flow process. : Time step, with a value of 1 hour, is the discrete interval of the time dimension in forward modeling, balancing computational efficiency and simulation accuracy. Predicted gas emission values, derived from forward modeling operators. Acting on the parameter vector to be inverted To obtain, that is . : The gas content grid parameter vector to be inverted, containing the gas content parameters to be determined for all grid nodes in the three-dimensional geological model.

[0106] S332: Prior Distribution of Spatial Variable Covariance

[0107] Furthermore, in step S3, differentiated spatial constraints are applied to the inversion parameters, specifically by constructing a spatially variable covariance prior distribution. This prior distribution is differentiated based on the spatial relationship between the grid nodes and the mining face: for grid nodes within the dynamic mining influence range, a first numerical range of prior variance is set; for grid nodes outside the influence range, a second numerical range of prior variance is set, and spatial smoothing constraints are applied to maintain regional continuity. The first numerical range is larger than the second numerical range, and the dynamic mining influence range is dynamically determined based on the real-time advance speed of the working face. The specific implementation is as follows:

[0108] A composite prior is used, consisting of Gaussian prior (spatial variable covariance) and L1 regularization (sparse constraints).

[0109]

[0110]

[0111] in, The spatially variable covariance prior distribution adopts a composite form of Gaussian prior + L1 regularization to constrain the reasonable distribution range of the parameters to be inverted. : Parameter vector to be inverted The transpose of the matrix; Spatial variational covariance matrix The inverse matrix, By combining a diagonal matrix (elements of which are...) ) and a discrete Laplace operator matrix (To achieve spatial smoothness) to construct, i.e. ,in The diagonal variance matrix (diagonal elements correspond to the grid nodes) or ), As a smoothing strength coefficient, in the influence radius The outer value is relatively large (e.g.) ),exist The inner value is relatively small (e.g.) ). : Sparsity constraint factor, range of values The method, determined through cross-validation, is used to promote the sparsity characteristics of the parameters to be inverted and suppress invalid fluctuations. : Parameter vector to be inverted The L1 norm is used to implement sparse constraints and improve model stability.

[0112] The core of this composite prior distribution is the combination of "spatial differentiation constraints" and "parameter sparsity constraints," which ensures both the accuracy of model reconstruction in the mining-affected area and suppresses invalid parameter fluctuations, thereby improving model stability. For spatially variable covariance matrices, their core function is to set differentiated prior constraint strengths based on whether grid nodes are within the range of mining influence.

[0113] (1) Radius of influence of mining The internal grid nodes are located in the area directly affected by the mining project and require high-precision dynamic reconstruction; therefore, a large prior variance is set. This allows for significant adjustments to the gas content parameters in the region during the inversion process to fully match real-time observation data;

[0114] (2) Radius of impact of mining External grid nodes: Not directly affected by mining operations, their gas occurrence is relatively stable, therefore a smaller prior variance is set. Simultaneously, a strong smoothing constraint is introduced through the spatial correlation matrix—even without real-time observation data, the spatial continuity of the model can be maintained through the parameter correlation between adjacent nodes, avoiding unfounded parameter mutations. The spatial correlation matrix is ​​used to determine the radius of influence of mining activities. The matrix with strong smoothing constraints applied to the outer nodes reduces parameter uncertainty in regions without observed data.

[0115] The radius of influence of mining activities is calculated using the following formula: It represents the spatial range by which mining operations affect the occurrence of coal seam gas. The mining advance speed is a real-time changing engineering parameter that reflects the progress of the mining operation. Time constant of mining impact, range of values This characterizes the temporal characteristics of the impact of mining activities extending outward from the mining work surface; Radius of impact of mining activities The prior variance of the internal grid nodes allows for a relatively large adjustment of the parameters in this area, ensuring high-precision local reconstruction; Radius of impact of mining activities The prior variance of the outer grid nodes limits the fluctuation range of parameters in this region and maintains the overall stability of the model;

[0116] S333: Likelihood Function

[0117] Assume the observed noise follows a Gaussian distribution:

[0118]

[0119] Likelihood function: assuming the observed noise follows a Gaussian distribution, used to quantify the parameters to be inverted. Corresponding predicted values ​​and observed data The degree of matching; The fused data vector is obtained by weighted fusion of gas emission data after multi-scale decomposition and serves as the input observation data for Bayesian inversion. Observation data vector The dimension, i.e., the total number of observed data; The noise covariance matrix, obtained by estimating the variance of different scale components, reflects the noise distribution characteristics of the observed data (variance of low-frequency components). ,high frequency ); Noise covariance matrix The determinant value; Noise covariance matrix The inverse matrix; variance of low-frequency components The upper limit of noise variance for low-frequency components is set because low-frequency data is highly stable, resulting in a lower noise tolerance; the variance of high-frequency components... The upper limit of noise variance for high-frequency components is set because high-frequency data is easily interfered with, resulting in a relatively high noise tolerance.

[0120] S334: Solving for the posterior probability

[0121] The constructed Bayesian inversion framework transforms the problem into an optimization problem by maximizing the posterior probability estimate. The Alternating Direction Multiplier Method (ADMM) is then used to numerically solve this problem. ADMM is a numerical algorithm for solving constrained optimization problems, specifically for efficiently solving composite optimization problems with L1 regularization, to obtain the optimal estimate of the gas content parameter.

[0122]

[0123] Iterative convergence condition of the alternating direction multiplier method: That is, when the relative change in the parameter vector between two consecutive iterations is less than When the iteration converges, the computation is stopped.

[0124] : Parameter vector to be inverted The optimal estimate is obtained by maximizing the posterior probability, which is the final inversion result of the gas content grid parameters. Weighted sum of squared residuals is equivalent to This characterizes the degree of deviation between the forward modeling predictions and the observed data; : No. The parameter vector to be inverted in the next iteration; : No. The parameter vector to be inverted in the next iteration.

[0125] In step S33 above, this invention introduces sparsity constraints and smooth background constraints to construct a multi-scale inversion framework based on Bayesian theory. It achieves differentiated constraints between mining-affected and non-affected areas through a spatially variable covariance prior matrix, balancing the characterization of local anomalies in gas occurrence with the preservation of spatial continuity. Furthermore, it establishes a closed-loop system of multi-scale data decomposition-Bayesian inversion-dynamic correction, integrating multi-source real-time monitoring data and geological structural information to achieve high-precision dynamic reconstruction of the gas occurrence state of complex coal seam groups, improving adaptability to complex structures and the dynamic impact of mining.

[0126] S34: Model Validation and Uncertainty Analysis

[0127] S341: Verification

[0128] The measured gas content data from the reserved verification boreholes were compared with the model output results, and the relative error was required to be ≤15%.

[0129] S342: Quantification of Uncertainty

[0130] The 95% confidence interval of gas content at each grid node is calculated based on the posterior covariance matrix to quantify the model's risk prediction.

[0131] In practical applications, considering that this method integrates multiple complex technical modules such as multi-scale geological modeling, dual-medium seepage simulation and multi-scale Bayesian inversion, and that field applications need to balance model accuracy, computational efficiency and engineering applicability, while avoiding the risks of parameter calibration difficulties and numerical instability that may be caused by one-time full module integration, this embodiment of the invention proposes a phased implementation strategy, which ensures the feasibility and reliability of the method through step-by-step verification and iterative optimization.

[0132] Phase 1 (Static High-Precision Modeling): Complete the static model construction in steps S1 to S2, calibrate parameters based on historical data, and achieve high-precision inversion of the static distribution of gas occurrence.

[0133] The second stage (dynamic module integration) involves adding the mining-induced damage coupled permeability model from step S22 (which dynamically characterizes the impact of coal seam fracture opening and closing on gas seepage during mining by associating mining-induced stress with fracture porosity and permeability through volumetric strain and damage variables) and the real-time data acquisition module to achieve dynamic evolution simulation.

[0134] Phase 3 (Optimization and Iteration): Based on feedback from field applications, optimize the weight allocation, inversion parameters, and model verification criteria in step S3 to improve adaptability to complex structural regions.

[0135] In summary, the present invention provides a dynamic reconstruction method for gas occurrence in complex coal seam groups based on spatiotemporal evolution field theory. This method constructs a geological model integrating structural curvature and fractal dimension to generate a permeability field matching the geological structure. It establishes a dual-medium gas spatiotemporal evolution model coupled with mining-induced damage as the physical kernel. Monitoring data is processed using multi-scale decomposition technology, and a Bayesian inversion framework incorporating spatially variable prior constraints is constructed for dynamic solution and closed-loop correction. Therefore, this invention achieves high-precision, real-time dynamic reconstruction of the gas occurrence state in coal seams with complex structures, effectively improving the accuracy of decision-making in mine gas disaster prevention and control.

[0136] like Figure 5 The schematic diagram illustrating the algorithm's signal reconstruction performance shows that, faced with a noisy synthetic signal, the reconstructed signal accurately matches the waveform trend of the original synthetic signal, demonstrating the powerful filtering capability of the multi-scale data decomposition technique (combining the MPSTFT algorithm and Ricker wavelet) against invalid interference. The amplitude difference between the reconstructed and synthetic signals is small, with no significant distortion, verifying the reliability of the subsequent Bayesian inversion framework. This indicates that the technique provides low-biased data support for dynamic reconstruction of gas occurrence, further ensuring the accuracy of gas occurrence status monitoring and reconstruction.

[0137] It should also be noted that, in the embodiments of this application, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0138] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined in the embodiments of this application may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown in this application, but is to be accorded the widest scope consistent with the principles and novel features disclosed in the embodiments of this application.

Claims

1. A method for dynamic reconstruction of gas occurrence in complex coal seam groups based on spatiotemporal evolution field theory, characterized in that, include: S1: Based on geological structural data, a three-dimensional geological model of the coal seam is established. By quantifying the curvature and interface complexity of the structure, an initial permeability distribution field of the coal seam that matches the structural space is generated. In step S1, generating the initial permeability distribution field includes: The curvature and interface complexity of the quantification structure are as follows: the structural curvature σ, which reflects the degree of structural bending, is calculated based on seismic interpretation data, and the fractal dimension D, which reflects the degree of irregularity, is calculated based on the structural interface data exposed by boreholes. Based on statistical regression methods, a permeability control model is established with the constructed curvature σ and fractal dimension D as variables: ,in, : Permeability control function, used to calculate the permeability distribution of coal seam grids, and the output value is the coal seam permeability of the corresponding grid node; : Basic value of original coal seam permeability; The initial permeability distribution field is generated by substituting the structural curvature σ and fractal dimension D at each location in space into the permeability control model K(σ,D). S2: Using the initial permeability distribution field as input, a spatiotemporal evolution model describing the occurrence and migration of gas in the matrix and fractures is established based on the dual-medium theory. The spatiotemporal evolution model is coupled with a dynamic damage mechanism driven by mining stress to correct fracture permeability in real time. In step S2, the spatiotemporal evolution model is a dynamic model based on the dual-medium theory. Its construction includes: dividing the coal seam into a coal matrix system and a fracture system, and establishing the following: the governing equations describing the free gas pressure diffusion in the fracture system; the governing equations describing the dynamics of adsorbed gas content in the coal matrix system; and the coupling relationship describing the gas mass exchange between the coal matrix system and the fracture system. S3: Collect real-time monitoring data of mine gas outbursts and extract their different frequency characteristic components using multi-scale signal decomposition technology; construct a Bayesian inversion framework using the spatiotemporal evolution model as the physical forward modeling constraint; wherein, the Bayesian inversion framework incorporates spatially variable prior information to implement differentiated spatial constraints on the inversion parameters of the mining-affected area and the non-affected area; by solving the framework, dynamically reconstruct the gas occurrence parameter field and use it to update the spatiotemporal evolution model to achieve closed-loop correction; Specifically, the differentiated spatial constraints on the inversion parameters of the mining-affected and non-affected areas are achieved by constructing a spatially variable covariance prior distribution. This prior distribution is set differently based on the spatial relationship between the grid nodes and the mining face: for grid nodes within the dynamic mining-affected area, a first numerical range of prior variance is set; for grid nodes outside the affected area, a second numerical range of prior variance is set, and spatial smoothing constraints are applied to maintain regional continuity. The first numerical range is greater than the second numerical range, and the dynamic mining-affected area is dynamically determined based on the real-time advance speed of the working face.

2. The method according to claim 1, characterized in that, In step S1, when constructing the three-dimensional geological model, tetrahedral meshes are used for subdivision, and different mesh sizes are set for densely structured areas and regular areas, wherein the mesh size of densely structured areas is smaller than that of regular areas.

3. The method according to claim 1, characterized in that, The spatiotemporal evolution model is coupled with the mining-induced damage mechanism to correct the permeability of the fracture system in real time according to the mining process, specifically including: Calculate the damage variables and volumetric strain of the coal body based on mining-induced stress; Based on the aforementioned damage variables and volumetric strain, the porosity of the fracture system is dynamically adjusted. According to the fracture cubic law, the permeability of the fracture system is updated in real time using the corrected porosity.

4. The method according to claim 1 or 3, characterized in that, The spatiotemporal evolution model is numerically simulated by simultaneously solving the following governing equations: Gas pressure diffusion equation for fractured systems based on gas pseudo-pressure linearization; The gas exchange rate equation between the matrix system and the fracture system based on the linear driving force assumption; Kinetic equation for adsorbed gas content within the matrix system; Equation of permeability evolution in fractured systems coupled with mining-induced damage.

5. The method according to claim 1, characterized in that, In step S3, different frequency feature components are extracted using multi-scale signal decomposition technology, specifically as follows: Based on the matching pursuit principle, the time series data of gas outburst is decomposed into multi-scale components, which are separated into different frequency components that represent large-scale background changes, meso-scale transitional changes and small-scale abnormal fluctuations. The decomposed components are then weighted and fused based on the weights determined by historical data to form a fused data vector for inversion.

6. The method according to claim 1, characterized in that, The Bayesian inversion framework constructed in step S3 has a posterior probability distribution maximization problem that is equivalent to the following optimization problem: in, To obtain the optimal estimate of the gas content parameters to be inverted, The forward modeling operator is constructed based on the aforementioned spatiotemporal evolution model. For the observed data vector, Fit residual terms to the weighted data; For spatial variation covariance matrix Prior constraints, is a sparse regularization term used to facilitate the characterization of local anomalies; m is the vector of parameters to be inverted; : Parameter vector to be inverted The transpose of .

7. The method according to claim 6, characterized in that, The optimization problem is solved numerically using the alternating direction multiplier method to obtain the optimal estimate of the gas content parameter.

Images