Ground surface and hole interior combined cavern hydrogen storage bank group geological condition detection method
By combining surface and borehole detection methods and utilizing Pearson correlation and Bayesian progressive inversion algorithms, the problems of multiple solutions and insufficient coverage in the geological exploration of cave hydrogen storage reservoirs were solved, and high-precision geological condition assessment of cave hydrogen storage reservoirs was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-20
- Publication Date
- 2026-04-03
AI Technical Summary
Existing technologies for geological exploration of cavern hydrogen storage facilities suffer from problems such as multiple solutions in the inversion results of a single geophysical method, lack of quantitative structural constraints for different geophysical methods, difficulty in identifying micro-fractures by surface exploration, and limited drilling coverage, leading to inaccurate exploration and misjudgment of site selection.
A combined surface and borehole detection method was adopted, and data were obtained through electromagnetic exploration and ground-penetrating radar. A joint inversion objective function with Pearson correlation constraints was constructed, and prior information constraints were established by combining borehole measured data. A Bayesian progressive inversion algorithm was used to construct a refined geological model.
It reduces the ambiguity of inversion, improves detection accuracy and reliability, can accurately assess the geological suitability of cavern hydrogen storage reservoirs, avoids misjudgment and insufficient coverage, and provides detailed geological data.
Smart Images
Figure CN121784840A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of geological condition detection, specifically to a method for detecting geological conditions of a rock cave hydrogen storage reservoir group that combines surface and borehole methods. Background Technology
[0002] As a crucial component of energy structure transformation, hydrogen energy's midstream "hydrogen storage" segment is of paramount importance. With the continuous growth of energy demand and the pursuit of clean energy, the application of hydrogen energy is becoming increasingly widespread. Cavern hydrogen storage facilities, due to their advantages such as large storage capacity, small footprint, and low long-term operating costs, have become an important development direction for large-scale hydrogen energy storage. They are of great significance for promoting the development of the hydrogen energy industry, achieving efficient energy storage and utilization, and promoting the optimization and upgrading of the energy structure. In the fields of geophysical exploration and geotechnical engineering detection technology, accurate detection of the geological conditions of cavern hydrogen storage facilities is key to ensuring their safe and stable operation.
[0003] Currently, geological exploration of underground caverns mainly relies on single or simple combinations of geophysical methods. Common methods include simple surface ground-penetrating radar (GPR), which obtains information about the subsurface medium by transmitting and receiving electromagnetic waves; high-density electrical resistivity tomography (EDT), which infers geological structure by measuring the resistivity at different locations underground. Electromagnetic methods are also used to understand subsurface geological conditions by studying the distribution and changes in electromagnetic fields. These methods can provide some data and information for geological exploration to a certain extent.
[0004] However, these existing technologies have significant drawbacks in practical applications. The inversion results of a single geophysical method (such as electromagnetic methods) often exhibit ambiguity, meaning different underground geological models may produce the same observational data, which can easily lead to misjudgments of geological structures such as faults and fracture zones. Furthermore, the lack of quantitative structural constraints between different geophysical methods (such as resistivity and dielectric constant) results in inconsistent spatial distributions of the inverted geological models, making them difficult to integrate. In addition, surface exploration struggles to accurately identify minute fissures around caves, while drilling, although accurate, only covers point-like areas and cannot provide comprehensive surface or volumetric exploration of the surrounding rock of caves. Summary of the Invention
[0005] The purpose of this invention is to overcome the above-mentioned technical deficiencies and propose a method for detecting the geological conditions of a rock cave hydrogen storage reservoir group that combines surface and borehole methods.
[0006] To achieve the above-mentioned technical objectives, the present invention adopts the following technical solution: This invention provides a method for detecting the geological conditions of a rock cave hydrogen storage reservoir group that combines surface and borehole methods, including: S1: Conduct electromagnetic exploration and ground-penetrating radar detection on the surface of the proposed cave hydrogen storage area to obtain resistivity observation data and first dielectric constant observation data; map the resistivity observation data and the first dielectric constant observation data into the same three-dimensional spatial grid, construct a joint inversion objective function with Pearson correlation constraint term, perform joint inversion, and obtain a first-generation geological model reflecting the structure of the underground medium; S2: Conduct ground-penetrating radar exploration inside the cave to obtain observation data on the second dielectric constant of the near-field surrounding rock; simultaneously, conduct drilling sampling and testing inside the cave to obtain rock mass physical property data at the borehole location; establish a mapping relationship between depth and physical property parameters based on the rock mass physical property data, and construct prior information constraint conditions containing model constraint vectors and confidence weight matrices accordingly. S3: Construct a Bayesian progressive inversion objective function that includes the prior information constraints, and use the prior information constraints to perform constrained inversion on the second dielectric constant observation data to obtain a fine geological model of the cave's near-field surrounding rock; based on the first-generation geological model and the fine geological model, comprehensively determine the geological suitability of the cave hydrogen storage reservoir.
[0007] In some embodiments, in step S1, the joint inversion objective function includes an electromagnetic inversion objective function. and ground-penetrating radar inversion objective function Their expressions are as follows: in, and These are the resistivity data fitting term and the ground-penetrating radar data fitting term, respectively. and These are the regularization constraints for the resistivity model and the dielectric constant model, respectively. For Pearson correlation constraints; This is the regularization factor.
[0008] In some embodiments, the Pearson correlation constraint term The construction method is as follows: in, For the first resistivity model parameters within each subdomain With dielectric constant model parameters The Pearson correlation coefficient; The These are weighting coefficients, and their values are determined based on the standard deviation of the parameters within the subdomain: when the resistivity standard deviation... Greater than the first threshold and the standard deviation of dielectric constant Greater than the second threshold hour, ,otherwise .
[0009] In some embodiments, step S1 involves mapping the resistivity observation data and the first dielectric constant observation data to the same three-dimensional spatial grid. Specifically, this includes: establishing a unified three-dimensional inversion grid, such that each grid node simultaneously possesses resistivity and dielectric constant attributes; and, considering the difference in resolution between electromagnetic exploration and ground-penetrating radar detection, using interpolation or grid refinement methods to align the two types of data to the center point of the three-dimensional inversion grid, so as to facilitate the calculation of the Pearson correlation coefficient of the local area grid.
[0010] In some embodiments, in step S3, the Bayesian progressive inversion objective function adopts a sequential update strategy, updating the inversion result once for each additional prior information of a borehole. For the Constraints introduced by each borehole, inverting the objective function Represented as: in, Indicates the first Phase-specific ground-penetrating radar observation data, Indicates the first The model parameters to be inverted at this stage For the orthogonal operator, Indicates the first The model constraint vector provided by each borehole Let be the covariance matrix of the borehole information structural constraints. and This is the regularization parameter.
[0011] In some embodiments, the model constraint vector in the prior information constraint conditions The specific construction process is as follows: For the Indoor experiments were conducted on rock cores obtained from each borehole to obtain measured values of lithology, fracture density, and dielectric constant at different depths. By employing polynomial fitting or discrete mapping, a numerical correspondence between the depth position on the borehole trajectory line and the measured value of the dielectric constant is established. The numerical correspondences are mapped to the spatial nodes in the inversion mesh that intersect with the borehole trajectory, and the measured dielectric constant values at these spatial nodes are used to form the model constraint vector. .
[0012] In some embodiments, the covariance matrix in the prior information constraints The specific construction process is as follows: Based on the spatial location of the borehole trajectory, a diagonal weighted matrix is constructed as the covariance matrix. ; exist In the model, diagonal elements corresponding to spatial nodes on the borehole trajectory are assigned high-confidence weight values, while elements corresponding to spatial nodes far from the borehole trajectory are assigned low-confidence weight values or zero values. The high-confidence weight values are greater than preset confidence weight values and are used to force the model parameters towards the model constraint vector during the inversion process. The low confidence weight value is less than the preset confidence weight value.
[0013] In some embodiments, in step S3, the inversion process is solved iteratively using the Gauss-Newton method, and the iteration termination condition is set as follows: When the Pearson correlation coefficient converges and the data fit difference is less than a preset threshold, the iteration stops and the final geological model is output; or, when the change in the objective function value between two consecutive iterations is less than a preset convergence tolerance, the iteration stops.
[0014] In some embodiments, in step S2, the arrangement of the drilling and sampling inside the cave is as follows: k equally spaced drilling locations are set along the axial direction of the tunnel at the front end of the rock cave storage, where k is 2 to 3; the drilling depth at each drilling location is 10 to 30 m to cover the effective area of the ground-penetrating radar. The ground-penetrating radar detection lines are laid out on the sidewalls, arch waists, and arch tops of the cave to obtain dielectric constant data of the surrounding rock within a 50m range.
[0015] In some embodiments, step S1 further includes a data preprocessing step before the joint inversion, the data preprocessing step including: A preliminary screening was conducted on historical geological and structural data, historical earthquake data, and tectonic activity information for the proposed reservoir area. If the proposed reservoir area is an abandoned mine shaft, historical mining data and the distribution of empty areas must be verified; the electromagnetic exploration lines and ground-penetrating radar lines will only be deployed in areas that pass the initial screening.
[0016] Compared with existing technologies, the beneficial effects of the geological condition detection method for cavern hydrogen storage reservoirs combining surface and borehole methods provided by this invention include: (1) During the surface exploration stage, a joint inversion objective function with Pearson correlation constraints is constructed to force the resistivity model obtained by the electromagnetic method and the dielectric constant model obtained by the ground-penetrating radar to maintain statistical consistency in the spatial structure variation trend. At the same time, the weighting coefficient is used to apply constraints only in areas where the physical property standard deviation exceeds the threshold. This mechanism overcomes the inherent defect of multiple solutions in the inversion results of a single geophysical method (such as resistivity or dielectric constant alone). It utilizes the coupling characteristics of the responses of two different geophysical fields to the boundary of the same geological body (such as fault fracture zone) to effectively eliminate false anomalies that may occur in a single method, making the geological model obtained by inversion clearer and more reliable on the structural boundary, thereby avoiding the error in hydrogen storage site selection due to misjudgment of geological structure.
[0017] (2) During the exploration phase inside the cave, prior information constraints based on borehole measurement data were constructed. Specifically, the measured "depth-physical property" data of the borehole cores were transformed into model constraint vectors, and a covariance matrix containing confidence weights was constructed. Using the Bayesian progressive inversion algorithm, the "hard data" of the borehole location was substituted as a strong constraint term into the ground-penetrating radar inversion equation. This method reduces the uncertainty of the inversion model and can more accurately quantify the integrity and water content of the surrounding rock, which is crucial for assessing whether hydrogen will escape.
[0018] (3) A tiered detection strategy is adopted: First, a large-scale scan is conducted using electromagnetic and radar joint inversion with a large spacing (20-50m) on the surface to quickly identify macroscopic geological structures; then, in the preliminarily qualified caverns, high-resolution wall-mounted radar detection and borehole (10-30m deep) sampling are used for microscopic detailed investigation. This combination of surface and borehole methods effectively compensates for the shortcomings of single detection methods, avoiding the problem that it is difficult to accurately identify deep micro-fractures by relying solely on surface detection, and solving the problem that relying solely on drilling has too small a coverage area and is prone to missing blind spots. This system can comprehensively identify the key geological elements affecting the stability of cavern hydrogen storage facilities, providing detailed and multi-scale geological basis for the site selection safety and subsequent support design of hydrogen storage facilities. Attached Figure Description
[0019] Figure 1 A flowchart of the geological exploration technology route provided for this application; Figure 2 A schematic diagram showing the layout of monitoring points for electromagnetic and ground-penetrating radar detection provided in this application; Figure 3 A schematic diagram showing the distribution of drilling points inside the cave provided in this application; Figure 4 A flowchart of the joint inversion process of electromagnetic and ground-penetrating radar detection provided for this application. Detailed Implementation
[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0021] This application mainly uses a combination of surface and borehole data to invert and explore the geology of caves, which achieves the effect of reducing the ambiguity of inversion and improving the detection accuracy to assess the geological suitability of cave hydrogen storage reservoirs. The following is a further detailed description of this application.
[0022] Example 1: Please refer to Figure 1 The geological condition detection method for cave hydrogen storage reservoirs combining surface and borehole data provided in this application includes a multi-source surface data acquisition and joint inversion stage, a cave interior data acquisition and prior information construction stage, and a precise inversion stage based on prior constraints. First, multi-source surface data is acquired and jointly inverted to obtain a first-generation geological model. Then, data is acquired inside the cave and prior information constraints are constructed. Finally, the prior information is used to constrain and invert the cave interior data to obtain a refined geological model. Based on these two models, the geological suitability of the cave hydrogen storage reservoir is comprehensively determined, achieving the effects of reducing inversion ambiguity, improving detection accuracy, and accurately assessing the geological conditions of the cave hydrogen storage reservoir. This is because by combining multi-source data and constraining inversion, the data information from both the surface and the cave interior is fully utilized, reducing the uncertainties brought about by single data and single inversion methods.
[0023] Specifically, the multi-source data acquisition and joint inversion stage of the Earth's surface includes two aspects: data acquisition and joint inversion.
[0024] Regarding data acquisition, data preprocessing is first performed, including initial screening of historical geological structure data, historical earthquake data, and tectonic activity information for the proposed reservoir area. If the proposed reservoir area is an abandoned mine, historical mining data and the distribution of empty areas need to be verified, and survey lines are only laid in areas that pass the initial screening. Subsequently, electromagnetic exploration and ground-penetrating radar detection are carried out on the surface of the proposed cavern hydrogen storage reservoir area. Figure 2 As shown, the specific layout method is as follows: multiple parallel survey lines are laid out above the proposed reservoir, with a survey line spacing of 20~50m and a point spacing of 5~15m. Figure 2 The image illustrates the relative positions of surface electromagnetic exploration points (diamond-shaped markers) and ground-penetrating radar (GPR) monitoring points (square markers). Electromagnetic exploration equipment utilizes specialized electromagnetic instruments, which transmit electromagnetic waves underground to measure the electromagnetic field response at different locations, thereby obtaining resistivity data. Ground-penetrating radar, on the other hand, uses high-frequency electromagnetic waves to detect the distribution of subsurface media, acquiring first dielectric constant data for the 0-50m shallow depth. Both electromagnetic instruments and GPR are mature geophysical exploration devices; however, other instruments with similar functions can be used as alternatives.
[0025] In terms of joint inversion, combined with Figure 4 The joint inversion process illustrated involves mapping resistivity and first dielectric constant observation data to the same three-dimensional spatial grid. First, a unified three-dimensional inversion grid is established, which functions like a three-dimensional coordinate system, with each grid node possessing both resistivity and dielectric constant attributes. To address the resolution differences between electromagnetic exploration and ground-penetrating radar (GPR) detection, interpolation or grid refinement methods are used to align the two sets of data to the center point of the three-dimensional inversion grid. For example, when the resolution difference between the two sets of data is significant, an interpolation algorithm can be used to estimate the value of the grid center point based on the known data points; alternatively, the grid can be refined to more accurately map the data to the grid nodes.
[0026] Then, a joint inversion objective function incorporating Pearson correlation constraints is constructed. For example... Figure 4 As shown in the flowchart, after obtaining the initial resistivity model and initial dielectric constant model Then, determine whether the data fit difference meets the threshold. If the conditions are met, Pearson correlation constraints are introduced for joint updates. The joint inversion objective function includes the electromagnetic inversion objective function. and ground-penetrating radar inversion objective function Their expressions are as follows: in, and These are resistivity data fitting terms and ground-penetrating radar data fitting terms, respectively, used to measure the degree of fit between the inversion model and the observation data; and These are regularization constraints for the resistivity model and the dielectric constant model, respectively, to ensure the smoothness and stability of the models.
[0027] The Pearson correlation constraint term is constructed as follows: in, For the first resistivity model parameters within each subdomain With dielectric constant model parameters The Pearson correlation coefficient is calculated using the following formula: In the formula, This represents the number of virtual grids within the subdomain. and They represent the first The resistivity and dielectric constant parameters of each virtual grid. and These represent the mean values of resistivity and dielectric constant within the subdomain, respectively.
[0028] These are weighting coefficients, and their values are determined based on the standard deviation of the parameters within the subdomain. When the resistivity standard deviation... Greater than the first threshold and the standard deviation of dielectric constant Greater than the second threshold hour, ,otherwise This weighting coefficient setting ensures that constraints are applied only in areas with significant geological anomalies, avoiding the introduction of noise into a uniform background field. By minimizing this joint inversion objective function and performing joint inversion, a first-generation geological model reflecting the subsurface structure can be obtained.
[0029] Specifically, the stage of data acquisition and prior information construction inside the cave includes data acquisition and prior information construction.
[0030] In terms of data acquisition, geological radar detection is carried out inside the cave, such as... Figure 2 As shown, ground-penetrating radar survey lines were laid out on the sidewalls (points 1, 2, 4, 5), arch waist, and arch crown (point 3) of the cave interior to obtain observation data of the second dielectric constant within a 50m range of the surrounding rock in the near field. Simultaneously, drilling and sampling were conducted inside the cave, such as... Figure 3 As shown, equidistant [structures] are arranged along the axial direction at the front end of the cavern storage tank. Drilling locations ( Figure 3 (Three drilling points are shown in the image). The value is set to 2 to 3, and the drilling depth at each drilling location is 10 to 30 meters to cover the effective area detected by ground-penetrating radar. A rotary drilling rig can be used to extract core samples from underground. During drilling, drilling television is used to record the development of fractures and filling materials in the well; and laboratory experiments are conducted on the extracted core samples to test their resistivity, dielectric constant, density, and other physical properties.
[0031] In terms of prior information construction, a mapping relationship between depth and physical property parameters is established based on the acquired rock mass physical property data, and prior information constraints including model constraint vectors and confidence weight matrices are constructed accordingly.
[0032] Model constraint vector The specific construction process is as follows: for the first Core samples obtained from each borehole were used for laboratory experiments. Professional rock property testing instruments, such as dielectric constant analyzers, were employed to obtain measured values of lithology, fracture density, and dielectric constant at different depths. A polynomial fitting or discrete mapping method was used to establish a numerical correspondence between depth positions along the borehole trajectory and the measured dielectric constant values. This numerical correspondence was then mapped to spatial nodes in the inversion mesh that intersect with the borehole trajectory, and the measured dielectric constant values at these spatial nodes were used to construct the model constraint vector. .
[0033] covariance matrix The specific construction process is as follows: Based on the spatial location of the borehole trajectory, a diagonal weighted matrix is constructed as the covariance matrix. .exist In the diagram, the diagonal elements corresponding to the spatial nodes on the borehole trajectory are assigned high-confidence weight values (e.g., values of 10 ... to This is used to force the model parameters to conform to the model constraint vector during the inversion process. Elements that are close to each other and correspond to spatial nodes that are far from the borehole trajectory are assigned low confidence weights or zero values.
[0034] Specifically, the precise inversion stage based on prior constraints includes constructing the inversion objective function and performing the inversion solution.
[0035] A Bayesian progressive inversion objective function incorporating prior information constraints is constructed, employing a sequential update strategy. The inversion result is updated once for each additional borehole's prior information. For the ... Constraints introduced by each borehole, inverting the objective function Represented as: in, Indicates the first Phase-specific ground-penetrating radar observation data, Indicates the first The model parameters to be inverted at this stage For the orthogonal operator, Indicates the first The model constraint vector provided by each borehole Let be the covariance matrix of the borehole information structural constraints. and This is the regularization parameter.
[0036] In the inversion solution, the Gauss-Newton method is used for iterative solution. The iteration termination condition is set as follows: when the Pearson correlation coefficient converges and the data fitting difference is less than a preset threshold, the iteration stops and the final geological model is output; or, when the change in the objective function value between two consecutive iterations is less than a preset convergence tolerance, the iteration stops. Through this iterative solution process, the second dielectric constant observation data is constrained and inverted using prior information constraints to obtain a detailed geological model of the cave's near-field surrounding rock. Finally, as shown... Figure 1 As shown in the flowchart, the geological suitability of the cave hydrogen storage reservoir is comprehensively determined based on the first-generation geological model and the refined geological model. For example, the suitability is determined by analyzing whether there are fracture development zones or water-bearing structures in the model that affect the safety of hydrogen storage.
[0037] The implementation principle of this embodiment is as follows: By combining multi-source surface data and cave interior data, and utilizing Pearson correlation constraints and Bayesian progressive inversion, this embodiment fully leverages the advantages of different data and inversion methods. Joint inversion of multi-source surface data reduces the ambiguity of single inversion methods and obtains preliminary geological structure information. The introduction of cave interior data and prior information further constrains the ground-penetrating radar inversion using borehole measurement data, improving the accuracy and reliability of the detection. The entire method, through explicit physical and statistical constraints, effectively reduces inversion uncertainty, accurately assesses the geological conditions of cave hydrogen storage facilities, and provides a scientific basis for the site selection and construction of cave hydrogen storage facilities. Compared with existing technologies, it represents a significant improvement in detection accuracy and reliability.
[0038] Example 2: The geological condition detection method for a combination of surface and borehole hydrogen storage reservoirs provided in this application includes the following steps: S1, Surface Multi-Source Data Acquisition and Joint Inversion Stage: First, data preprocessing is performed, with initial screening of historical geological structure data, historical earthquakes, and tectonic activity information for the proposed reservoir area. If the proposed reservoir area is an abandoned mine, historical mining data and the distribution of empty areas need to be verified. Electromagnetic exploration lines and ground-penetrating radar lines are only deployed in areas that pass the initial screening. Then, electromagnetic exploration and ground-penetrating radar detection are carried out on the surface of the proposed cavern hydrogen storage reservoir area to obtain resistivity and first dielectric constant observation data. During this process, the preferred spacing between electromagnetic exploration lines is 20-50m, and the preferred point spacing is 5-15m. The resistivity and first dielectric constant observation data are uniformly mapped to the same three-dimensional spatial grid, and a joint inversion objective function incorporating Pearson correlation constraints is constructed. Joint inversion is then performed to obtain a first-generation geological model reflecting the subsurface medium structure.
[0039] S2, Cavern Interior Data Acquisition and Prior Information Construction Stage: Ground-penetrating radar (GPR) exploration is conducted inside the cavern to acquire observational data on the second dielectric constant of the near-field surrounding rock. Simultaneously, drilling and sampling are performed within the cavern to obtain rock mass physical property data at the borehole locations. Based on the rock mass physical property data, a mapping relationship between depth and physical property parameters is established, and prior information constraints, including model constraint vectors and confidence weight matrices, are constructed accordingly. These prior information constraints are implemented by converting the measured point data from the boreholes into hard constraint data in the inversion grid, ensuring the accuracy of the geological model at the borehole locations.
[0040] S3, Precise Inversion Stage Based on Prior Constraints: A Bayesian progressive inversion objective function incorporating prior information constraints is constructed. The prior information constraints are used to perform constrained inversion on the second dielectric constant observation data, obtaining a refined geological model of the cave's near-field surrounding rock. Based on the first-generation geological model and the refined geological model, the geological suitability of the cave hydrogen storage reservoir is comprehensively determined.
[0041] The implementation principle of this embodiment is as follows: The entire detection method of this embodiment has a clear process, and each step is closely linked. The data preprocessing step can eliminate some areas that are not suitable for the construction of cavern hydrogen storage facilities, improving the targeting and effectiveness of subsequent detection. The combination of joint inversion of multi-source surface data and prior-constrained inversion of data from inside the cave fully utilizes data from different locations and types, reduces the ambiguity of the inversion, and improves the detection accuracy. Finally, by synthesizing the geological model obtained from the two stages, the geological suitability of cavern hydrogen storage facilities can be accurately assessed, providing a scientific and reliable basis for the construction of cavern hydrogen storage facilities, which is a significant improvement compared to existing single or simple combination detection methods.
[0042] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.
Claims
1. A method for detecting geological conditions of a rock cave hydrogen storage reservoir group combining surface and borehole methods, characterized in that, include: S1: Conduct electromagnetic exploration and ground-penetrating radar detection on the surface of the proposed cave hydrogen storage area to obtain resistivity observation data and first dielectric constant observation data; map the resistivity observation data and the first dielectric constant observation data into the same three-dimensional spatial grid, construct a joint inversion objective function with Pearson correlation constraint term, perform joint inversion, and obtain a first-generation geological model reflecting the structure of the underground medium; S2: Conduct ground-penetrating radar exploration inside the cave to obtain observation data on the second dielectric constant of the near-field surrounding rock; simultaneously, conduct drilling sampling and testing inside the cave to obtain rock mass physical property data at the borehole location; establish a mapping relationship between depth and physical property parameters based on the rock mass physical property data, and construct prior information constraint conditions containing model constraint vectors and confidence weight matrices accordingly. S3: Construct a Bayesian progressive inversion objective function that includes the prior information constraints, and use the prior information constraints to perform constrained inversion on the second dielectric constant observation data to obtain a fine geological model of the cave's near-field surrounding rock; based on the first-generation geological model and the fine geological model, comprehensively determine the geological suitability of the cave hydrogen storage reservoir.
2. The method for detecting geological conditions of a rock cave hydrogen storage reservoir group combining surface and borehole methods according to claim 1, characterized in that, In step S1, the joint inversion objective function includes the electromagnetic inversion objective function. and ground-penetrating radar inversion objective function Their expressions are as follows: in, and These are the resistivity data fitting term and the ground-penetrating radar data fitting term, respectively. and These are the regularization constraints for the resistivity model and the dielectric constant model, respectively. For Pearson correlation constraints; This is the regularization factor.
3. The method for detecting geological conditions of a rock cave hydrogen storage reservoir group combining surface and borehole methods according to claim 1, characterized in that, The Pearson correlation constraint The construction method is as follows: in, For the first resistivity model parameters within each subdomain With dielectric constant model parameters The Pearson correlation coefficient; The These are weighting coefficients, and their values are determined based on the standard deviation of the parameters within the subdomain: when the resistivity standard deviation... Greater than the first threshold and the standard deviation of dielectric constant Greater than the second threshold hour, ,otherwise .
4. The method for detecting geological conditions of a rock cave hydrogen storage reservoir group combining surface and borehole methods according to claim 1, characterized in that, In step S1, the resistivity observation data and the first dielectric constant observation data are uniformly mapped to the same three-dimensional spatial grid. Specifically, this includes: establishing a unified three-dimensional inversion grid so that each grid node has both resistivity and dielectric constant attributes; and, considering the difference in resolution between electromagnetic exploration and ground-penetrating radar, using interpolation or grid refinement methods to align the two types of data to the center point of the three-dimensional inversion grid, so as to facilitate the calculation of the Pearson correlation coefficient of the local area grid.
5. The method for detecting geological conditions of a rock cave hydrogen storage reservoir group combining surface and borehole methods according to claim 1, characterized in that, In step S3, the Bayesian progressive inversion objective function adopts a sequential update strategy, updating the inversion result once for each additional prior information of a borehole. For the Constraints introduced by each borehole, inverting the objective function Represented as: in, Indicates the first Ground-penetrating radar observation data at different stages, Indicates the first The model parameters to be inverted at this stage For the orthogonal operator, Indicates the first The model constraint vector provided by each borehole Let be the covariance matrix of the borehole information structural constraints. and This is the regularization parameter.
6. The method for detecting geological conditions of a rock cave hydrogen storage reservoir group combining surface and borehole methods according to claim 5, characterized in that, The model constraint vector in the prior information constraint conditions The specific construction process is as follows: For the first Indoor experiments were conducted on rock cores obtained from each borehole to obtain measured values of lithology, fracture density, and dielectric constant at different depths. By employing polynomial fitting or discrete mapping, a numerical correspondence between the depth position on the borehole trajectory line and the measured value of the dielectric constant is established. The numerical correspondences are mapped to the spatial nodes in the inversion mesh that intersect with the borehole trajectory, and the measured dielectric constant values at these spatial nodes are used to form the model constraint vector. .
7. The method for detecting geological conditions of a rock cave hydrogen storage reservoir group combining surface and borehole methods according to claim 5, characterized in that, The covariance matrix in the prior information constraint The specific construction process is as follows: Based on the spatial location of the borehole trajectory, a diagonal weighted matrix is constructed as the covariance matrix. ; exist In the model, diagonal elements corresponding to spatial nodes on the borehole trajectory are assigned high-confidence weight values, while elements corresponding to spatial nodes far from the borehole trajectory are assigned low-confidence weight values or zero values. The high-confidence weight values are greater than preset confidence weight values and are used to force the model parameters towards the model constraint vector during the inversion process. The low confidence weight value is less than the preset confidence weight value.
8. The method for detecting geological conditions of a rock cave hydrogen storage reservoir group combining surface and borehole methods according to claim 1, characterized in that, In step S3, the inversion process is solved iteratively using the Gauss-Newton method, and the iteration termination condition is set as follows: When the Pearson correlation coefficient converges and the data fit difference is less than a preset threshold, the iteration stops and the final geological model is output; or, when the change in the objective function value between two consecutive iterations is less than a preset convergence tolerance, the iteration stops.
9. The method for detecting geological conditions of a rock cave hydrogen storage reservoir group combining surface and borehole methods according to claim 1, characterized in that, In step S2, the layout of the drilling and sampling inside the cave is as follows: k equally spaced drilling locations are set along the axial direction of the tunnel at the front end of the rock cave storage, where k is 2 to 3; the drilling depth at each drilling location is 10 to 30 m to cover the effective area of the ground-penetrating radar. The ground-penetrating radar detection lines are laid out on the sidewalls, arch waists, and arch tops of the cave to obtain dielectric constant data of the surrounding rock within a 50m range.
10. The method for detecting geological conditions of a rock cave hydrogen storage reservoir group combining surface and borehole methods according to claim 1, characterized in that, In step S1, a data preprocessing step is included before the joint inversion. The data preprocessing step includes: A preliminary screening was conducted on historical geological and structural data, historical earthquake data, and tectonic activity information for the proposed reservoir area. If the proposed reservoir area is an abandoned mine shaft, historical mining data and the distribution of empty areas must be verified; the electromagnetic exploration lines and ground-penetrating radar lines will only be deployed in areas that pass the initial screening.