Salt cavern hydrogen storage bank group geologic model construction method based on multi-means technology fusion

Abstract

The invention discloses a salt cavern hydrogen storage reservoir group geologic model construction method based on multi-means technology fusion. The method comprises the steps that a preliminary model is obtained through electromagnetic and earthquake independent inversion; performing primary joint inversion based on cross gradient constraint, and constructing a macroscopically consistent model pair; constructing a multi-physics field database through drilling sampling; and finally, by using drilling data as strong prior information, constructing a well-seismic-electric constraint term containing a rock physical statistical mapping relation and a formation interface geometric trend, carrying out constrained secondary inversion, and outputting a final three-dimensional geologic model. According to the technical scheme, through combination of two-stage inversion and strong geometric constraint, the problem of multiplicity of solutions of geophysical inversion is solved, one-dimensional well hole actual measurement data is expanded into three-dimensional space constraint by using a strong constraint term, the reliability of a geologic model in a well-free area is improved, and the recognition capability of a hydrogen storage bank group stratum boundary is enhanced.

Description

A Method for Constructing Geological Models of Salt Cavern Hydrogen Storage Clusters Based on Multi-Technology Fusion Technical Field

[0001] This invention relates to the field of geological modeling technology, specifically to a method for constructing a geological model of a salt cavern hydrogen storage reservoir group based on the fusion of multiple technologies. Background Technology

[0002] Salt caverns are widely recognized as ideal large-scale underground hydrogen storage sites due to their excellent creep properties, extremely low permeability, and self-healing capabilities. However, the extremely small diameter of hydrogen molecules, coupled with their strong diffusivity and permeability, places more stringent demands on the site selection, surrounding rock sealing evaluation, and detailed geological structure characterization of salt cavern hydrogen storage facilities compared to traditional natural gas storage facilities. Constructing a high-precision, multi-physics parameter-unified three-dimensional geological model is a prerequisite for ensuring the safe construction and operation of salt cavern hydrogen storage facilities.

[0003] Currently, geological exploration of salt caverns mainly relies on surface geophysical methods, but single methods have inherent limitations in terms of detection accuracy and reliability. For example, magnetotelluric (MT) methods are relatively sensitive to low-resistivity underground bodies (such as brine pockets and fracture zones), but their volume effect is significant, and their lateral and vertical resolution is low, making it difficult to accurately characterize rock layer boundaries. Although seismic exploration has high resolution for stratigraphic impedance interfaces, within salt rock bodies, due to the homogeneity and high-velocity characteristics of salt rock, energy shielding effects often occur, leading to a decrease in signal-to-noise ratio. Furthermore, single seismic inversion suffers from severe ambiguity problems when lacking constraints.

[0004] To overcome the limitations of single methods, multi-geophysical field joint inversion techniques have emerged. Existing joint inversion methods mainly utilize cross-gradient theory to find structural similarities between different physical properties (such as resistivity and velocity). However, existing joint inversion techniques still face the following technical bottlenecks in practical applications: First, most existing joint inversions only rely on the mutual constraints of geophysical data, lacking effective integration of borehole measured data (such as core sampling and logging curves). Although drilling data has extremely high accuracy, it is usually regarded as discrete verification points, failing to deeply integrate the rock physical statistical laws it contains as prior information into the inversion equations. This results in numerical deviations between the inverted physical parameters (resistivity, wave velocity) and the actual formation conditions.

[0005] Secondly, traditional well-seismic combined modeling methods typically only utilize well logging data for one-dimensional calibration or simple interpolation and extrapolation, making it difficult to effectively control the geometric morphology of formation interfaces far from the wellbore in three-dimensional space. In complex geological contexts, mathematical interpolation alone cannot accurately reflect the sedimentary patterns and tectonic trends of the formations, leading to distortions in the structural morphology of the constructed geological model in well-free areas. This makes it difficult to meet the micron-level accuracy evaluation requirements for geological boundaries in salt cavern hydrogen storage reservoirs.

[0006] Therefore, there is an urgent need for a geological model construction method for salt cavern hydrogen storage reservoir groups that can deeply integrate electromagnetic, seismic, and in-situ drilling test information, overcome the limitations of single methods through mathematical and physical constraint mechanisms, and effectively utilize well data to constrain the three-dimensional spatial geometry. Summary of the Invention

[0007] The purpose of this invention is to overcome the above-mentioned technical deficiencies and propose a method for constructing a geological model of a salt cavern hydrogen storage reservoir group based on the integration of multiple technologies.

[0008] To achieve the above technical objectives, the present invention adopts the following technical solution: The present invention provides a method for constructing a geological model of a salt cavern hydrogen storage reservoir group based on the fusion of multiple technologies, including: S1: organizing historical geological data of the target area, conducting preliminary screening for reservoir suitability, and delineating areas with concentrated salt mineral resources; scanning the delineated area using electromagnetic exploration technology, and performing inversion based on electromagnetic data to obtain a preliminary resistivity inversion model; S2: deploying a three-dimensional seismic detection system based on the electromagnetic exploration results, collecting seismic wave data and performing independent inversion to obtain a preliminary seismic wave velocity inversion model; S3: performing grid resampling and matching, using the preliminary resistivity inversion model of step S1 and the preliminary seismic wave velocity inversion model of step S2 as initial guess models, and performing cross-gradient constraint-based matching between electromagnetic data and seismic data. The initial joint inversion is performed to construct a joint inversion model pair that includes macroscopic structural consistency; S4: Based on the interpretation results of the joint inversion model pair, the proposed reservoir area is selected for drilling and sampling, and core and formation fluid samples are collected; rock mechanics experiments, porosity and permeability tests and physical property tests are conducted to construct a depth-rock mechanics parameter-resistivity-wave velocity multiphysics database and a depth-rock interface database; S5: Using the lithological data obtained from drilling as strong prior information, a well-seismic-electric constraint term including rock physical statistical mapping relationship and formation interface geometric trend is constructed; the joint inversion model pair output in step S3 is used as the starting model for the secondary inversion, and the well-seismic-electric constraint term is introduced to construct a constrained secondary inversion objective function, and constrained secondary inversion is performed to output the final three-dimensional geological model of the salt cavern hydrogen reservoir group.

[0009] In some embodiments, in step S1, the electromagnetic data inversion employs a regularized Gauss-Newton inversion method to logarithmize the subsurface resistivity parameters, denoted as... Define the electromagnetic inversion objective function. Fitting terms from electromagnetic inversion data and electromagnetic inversion regularization term composition: The electromagnetic inversion data fitting term is defined as follows: In the formula For electromagnetic data weighting matrix, This is a vector of measured electromagnetic observation data. Here, we define the electromagnetic forward modeling function for calculating theoretical electromagnetic response data based on the resistivity model; the electromagnetic inversion regularization term is defined as follows: In the formula This is the discretization matrix of the three-dimensional spatial gradient operator; Regularization factor; model parameters The iterative update formula is: in, The electromagnetic forward modeling function Regarding model parameters The Jacobian matrix.

[0010] In some embodiments, in step S2, the initial seismic wave velocity model is constructed using an optimization algorithm based on L-BFGS, and the seismic wave velocity parameters are logarithmically processed. Define the seismic inversion objective function. Fitting terms from seismic inversion data and earthquake inversion regularization term composition: Among them, the seismic inversion data fitting term is defined as In the formula Weighted matrix for earthquake data, This is a seismic forward modeling operator for calculating theoretical seismic wavefield data based on a velocity model. For seismic observation data; the seismic inversion regularization term is defined as follows: Here is the regularization factor for the velocity model; its iterative formula is: in: Step size factor It is an approximate Hessian inverse matrix. The gradient of the objective function.

[0011] In some embodiments, the specific steps of the joint inversion in step S3 include: resampling the electromagnetic inversion grid onto the seismic inversion grid to establish a unified inversion grid system; and constructing the joint inversion objective function. : in, and These are the fitting terms for electromagnetic inversion data and the fitting terms for seismic inversion data, respectively. and These are the electromagnetic inversion regularization term and the seismic inversion regularization term, respectively; The discretized cross gradient constraint term is defined as follows: In the formula The total number of effective grid cells, and For the first Gradient vectors at each grid point; to These are the weighting coefficients.

[0012] In some embodiments, in step S4, the criteria for selecting the proposed reservoir area are: based on the geological interpretation of the joint inversion model, the area is inferred to have a salt layer burial depth between 800 and 1500 m and a pure salt rock layer accounting for more than 80%; the test content includes porosity and permeability testing and biochemical suitability evaluation.

[0013] In some embodiments, in step S5, statistical mapping constraints are constructed using the lithological data obtained from drilling. and interface trend constraints The statistical mapping constraint term The discretized expression is: in, and The first The velocity and resistivity of each grid cell; The empirical regression function is pre-fitted based on the database described in step S4. The fixed coefficient vector that does not participate in the inversion iteration; the interface trend constraint term The construction method is as follows: (1) Perform tensor analysis on the joint inversion model output in step S3, calculate the gradient outer product matrix for each grid point and perform eigenvalue decomposition, and select the eigenvector corresponding to the largest eigenvalue as the formation normal vector of that point. (2) Calculate the normal gradient at the wellbore trajectory based on the logging data, and construct a normal gradient trend field covering the entire area using a spatial interpolation algorithm. subscript (3) Construct constraint formulas: ,in: For the set of inverted grid nodes; Represents model parameters At the node gradient vector at point With the normal vector of the strata at that location The dot product; These are the normalized weighting coefficients.

[0014] In some embodiments, in step S5, the constrained quadratic inversion objective function for: in, and These are the fitting terms for electromagnetic inversion data and the fitting terms for seismic inversion data, respectively. For discretized cross-gradient constraint terms; For statistical mapping constraints; and These are the interface trend constraints for the resistivity model and the velocity model, respectively. Weighting coefficients to maintain data consistency and structural coupling. Weighting coefficients are used to control the strength of prior constraints.

[0015] In some embodiments, the method for constructing the initial guessing model in step S3 further includes: using historical geological profiles around the target area to create a high-resolution geological model as a label; performing forward modeling and smoothing on the label to generate a low-resolution physical parameter model as input; constructing and training a structure recovery convolutional neural network; inputting the preliminary resistivity inversion model obtained in step S1 and the preliminary seismic wave velocity inversion model obtained in step S2 as input data into the trained neural network to predict a physical parameter model containing fine structural features; and using the predicted model as the initial guessing model for the primary joint inversion in step S3.

[0016] In some embodiments, the method for constructing a geological model of a salt cavern hydrogen storage reservoir group based on the fusion of multiple technologies further includes a hydrogen sealing performance evaluation step based on the final three-dimensional geological model. The hydrogen sealing performance evaluation step includes: conducting a biochemical suitability evaluation based on the formation fluid samples collected in step S4, analyzing the content of sulfate-reducing bacteria and hydrogen consumption reaction parameters; and constructing a three-dimensional geostress field based on the final model to simulate the hydrogen migration path and generate a risk cloud map.

[0017] In some embodiments, the method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies further includes a four-dimensional dynamic geological monitoring step during the operation of the hydrogen storage reservoir. The four-dimensional dynamic geological monitoring step during the operation of the hydrogen storage reservoir includes performing differential inversion on time-shifted electromagnetic and microseismic data collected at different time steps based on the final geological model, calculating the rate of change of formation parameters, and identifying salt cavern creep and hydrogen diffusion range.

[0018] Compared with existing technologies, the beneficial effects of the method for constructing a geological model of a salt cavern hydrogen reservoir group based on the fusion of multiple technologies provided by this invention are as follows: (1) It adopts a two-stage modeling process that combines "primary joint inversion" and "constrained secondary inversion". First, a joint inversion model pair with macroscopic structural consistency is constructed using the cross-gradient constraints of electromagnetic and seismic data. Then, using this as the starting model, well-seismic-electric constraint terms (including rock physical statistical mapping relationships and formation interface geometric trends) constructed using drilling data are introduced in the secondary inversion stage. This technique uses the results of primary inversion to provide a good initial structural framework for subsequent steps, avoiding the boundary distortion that may occur when directly using sparse well data for interpolation modeling. At the same time, the secondary inversion quantitatively corrects the model by introducing drilling measured data as strong prior information, thereby ensuring that the model conforms to the macroscopic geological structure, improving the consistency between the model's physical parameter values ​​and the measured data, and reducing the ambiguity of single geophysical inversion.

[0019] (2) The formation normal vector of the joint inversion model was extracted by constructing tensor analysis, and the normal gradient at the well trajectory was extended into a normal gradient trend field covering the entire area by combining spatial interpolation algorithm, thus constructing the interface trend constraint term. This technical feature extends the one-dimensional logging information that traditionally only exists in the well trajectory into a three-dimensional geometric trend constraint that can control the entire field. This allows the constructed geological model to maintain a layered morphology that conforms to the geological sedimentary laws or structural characteristics even in areas far from the well, thus improving the problem of unnatural model morphology that easily occurs in areas with complex geological structures using pure mathematical interpolation methods.

[0020] (3) By using discretized cross-gradient constraints, electromagnetic data, which is more sensitive to fluids and low-resistivity bodies, is jointly inverted with seismic data, which is more sensitive to the impedance interface of the formation waves. This technique forces the gradient vectors of the resistivity model and the velocity model to tend to be parallel or opposite during the iteration process, thereby achieving spatial consistency between the two at the geological structure boundary. This coupling mechanism uses the structural resolution capability of seismic data to compensate for the resolution deficiency caused by the volume effect of the electromagnetic method, and at the same time uses electromagnetic data to supplement the signal loss in the homogeneous zone or energy shielding zone inside the salt rock during seismic exploration, so that the final model can more clearly present the physical property interface between the salt rock and the surrounding rock. Attached Figure Description

[0021] Figure 1 is a flowchart illustrating the method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies provided in this embodiment of the invention.

[0022] Figure 2 is an iterative flowchart of the joint inversion algorithm in an embodiment of the present invention. Detailed Implementation

[0023] 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.

[0024] Referring to Figures 1 and 2, this embodiment provides a method for constructing a geological model of a salt cavern hydrogen storage reservoir cluster based on the fusion of multiple technologies. This method systematically integrates electromagnetic exploration, seismic exploration, and drilling logging data, utilizing cross-gradient constraints and well-seismic-electrical-intensity constraints to construct a high-precision three-dimensional geological model step by step. Specifically, the method includes the following steps: S1: Organizing historical geological data of the target area, conducting preliminary screening for reservoir suitability, and delineating concentrated salt mineral resource areas; scanning the delineated area using electromagnetic exploration technology, and performing inversion based on the electromagnetic data to obtain a preliminary resistivity inversion model.

[0025] In the initial phase of the project, historical geological data of the target area were first compiled to conduct preliminary screening for database suitability and delineate areas with relatively concentrated salt resources. Subsequently, electromagnetic exploration techniques (such as magnetotellurics (MT) or controlled-source audio-frequency magnetotellurics (CSAMT)) were used to scan the delineated areas.

[0026] After obtaining the measured electromagnetic data, an inversion process based on the electromagnetic data is performed to obtain a preliminary resistivity inversion model. To overcome the ambiguity of electromagnetic inversion and improve its stability, this embodiment employs a regularized Gauss-Newton inversion method. Before inversion, considering that the variation range of subsurface resistivity parameters typically spans multiple orders of magnitude, the subsurface resistivity parameters are logarithmically normalized to improve the condition number of the inversion problem. The model parameters are then denoted as... ,in is the resistivity.

[0027] The electromagnetic inversion objective function constructed in this step It consists of two parts: one part is the electromagnetic inversion data fitting term. The other part is the electromagnetic inversion regularization term. Its mathematical expression is: The electromagnetic inversion data fitting term is defined as follows: In the formula, The electromagnetic data weighting matrix is ​​usually taken as the inverse square root of the data covariance matrix, and is used to weight data of different qualities. This is a vector of measured electromagnetic observation data; This is the electromagnetic forward modeling function for calculating theoretical electromagnetic response data based on the resistivity model. The electromagnetic inversion regularization term is defined as follows: In the formula, This is the discretization matrix of the three-dimensional spatial gradient operator, used to constrain the smoothness of the model; This is a regularization factor used to balance the degree of data fit with the smoothness of the model.

[0028] Model parameters The iterative update is achieved by solving a system of linearized equations, as shown in the following formula: in, The electromagnetic forward modeling function Regarding model parameters The Jacobian matrix represents the rate of change of the data response caused by small changes in the model. Through the above iterative process, a preliminary resistivity model reflecting the underground electrical structure can be quickly obtained, effectively identifying low-resistivity anomaly zones (such as brine pockets) and high-resistivity salt rock masses.

[0029] S2: Based on the electromagnetic exploration results, a three-dimensional seismic detection system is deployed to collect seismic wave data and perform independent inversion to obtain a preliminary seismic wave velocity inversion model.

[0030] Based on the inversion results of electromagnetic exploration, a three-dimensional seismic detection system was deployed in the identified main salt rock area to collect high-density seismic wave data. Subsequently, the seismic data was independently inverted to obtain a preliminary seismic wave velocity inversion model.

[0031] In this step, the initial seismic velocity model is constructed using an optimization algorithm based on L-BFGS (Limited-memory Broyden–Fletcher–Goldfarb–Shanno). Similarly, to ensure the physical meaning and numerical stability of the parameters, the seismic velocity parameters are logarithmically transformed, denoted as... ,in This represents the longitudinal wave velocity.

[0032] Defined seismic inversion objective function Similarly, the fitting term from the seismic inversion data and earthquake inversion regularization term Composition, the expression is: Among them, the seismic inversion data fitting term is defined as In the formula, Weighting matrix for earthquake data; This is a seismic forward modeling operator for calculating theoretical seismic wavefield data based on a velocity model. For seismic observation data. The seismic inversion regularization term is defined as follows: , This is the regularization factor for the velocity model.

[0033] The L-BFGS algorithm does not require explicit storage of the Hessian matrix; its iterative formula is as follows: in, The step size factor is determined by the line search strategy; The approximate Hessian inverse matrix updated using the L-BFGS algorithm; The gradient of the objective function. This step utilizes the high-resolution characteristics of seismic data to clearly characterize stratigraphic horizons and fault structures.

[0034] S3: Perform grid resampling and matching. Using the preliminary resistivity inversion model from step S1 and the preliminary seismic wave velocity inversion model from step S2 as the initial guess model, perform a primary joint inversion of electromagnetic data and seismic data based on cross-gradient constraints to construct a joint inversion model pair that includes macroscopic structural consistency.

[0035] Please refer to Figure 2. Since electromagnetic inversion and seismic inversion usually use different grid partitioning strategies (sparse electromagnetic grid and dense seismic grid), before joint inversion, grid resampling and matching are first performed to map the two models to the same set of fine three-dimensional inversion grid system.

[0036] Subsequently, using the preliminary resistivity inversion model obtained in step S1 and the preliminary seismic wave velocity inversion model obtained in step S2 as initial guess models, a primary joint inversion based on cross-gradient constraints is performed on the electromagnetic data and seismic data. This step involves constructing a joint inversion objective function. To achieve: In this objective function, and These are the electromagnetic inversion data fitting term and the seismic inversion data fitting term as defined above; and These are the electromagnetic inversion regularization term and the seismic inversion regularization term as defined above.

[0037] In particular, The discretized cross-gradient constraint term is defined as follows: In the formula The total number of effective grid cells, and No. Resistivity gradient vector and velocity gradient vector at each grid point; to where represents the weighting coefficients for each term. By minimizing the cross gradient term, the cross product of the resistivity gradient and the velocity gradient is forced to approach zero, requiring that their directions of change remain parallel or opposite. This ensures that the geological structural boundaries (such as stratigraphic interfaces) reflected by the two remain spatially consistent even when physical properties differ, thus constructing a joint inversion model pair that includes macroscopic structural consistency.

[0038] In this step, to further improve the quality of the initial model, a deep learning-assisted initialization strategy can be introduced. Specifically, a high-resolution geological model is created using historical geological profiles surrounding the target area as labels. Forward modeling and smoothing are performed on these labels to generate a low-resolution physical parameter model as input. A structure recovery convolutional neural network is then constructed and trained. The preliminary inversion model obtained in steps S1 and S2 is fed into this trained network to predict a physical parameter model containing fine structural features. This predicted model is then used as the initial guessing model for the primary joint inversion, thereby accelerating convergence and reducing the risk of getting trapped in local minima.

[0039] S4: Based on the interpretation results of the joint inversion model, select the proposed reservoir area for drilling and sampling, collect rock cores and formation fluid samples; conduct rock mechanics experiments, porosity and permeability tests and physical property tests, and construct a depth-rock mechanics parameter-resistivity-wave velocity multiphysics database and a depth-rock interface database.

[0040] Based on the geological interpretation results of the aforementioned joint inversion model, drilling and sampling were conducted in the proposed reservoir area. The site selection criteria are typically: areas where the inferred salt layer depth is between 800 and 1500 m and the proportion of pure salt rock layers exceeds 80%.

[0041] During drilling, core samples and formation fluid samples were collected, and a series of laboratory experiments and tests were conducted. The tests mainly included: rock mechanics experiments (obtaining Young's modulus, Poisson's ratio, etc.), porosity and permeability tests (obtaining porosity and permeability), physical property tests (obtaining precise resistivity and wave velocity values), and biochemical suitability evaluation (analyzing sulfate-reducing bacteria content and hydrogen consumption reaction kinetic parameters). Based on the test results, a multiphysics database containing the correspondence between depth, rock mechanics parameters, resistivity, and wave velocity, as well as a depth-stratum interface database based on well logging layers, were constructed.

[0042] S5: Using the lithological data obtained from drilling as strong prior information, construct a well-seismic-electric constraint term that includes rock physical statistical mapping relationships and formation interface geometric trends; using the joint inversion model output in step S3 as the starting model for the secondary inversion, introduce the well-seismic-electric constraint term to construct a constrained secondary inversion objective function, perform constrained secondary inversion, and output the final three-dimensional geological model of the salt cavern hydrogen reservoir group.

[0043] Using lithological data obtained from drilling as strong prior information, a well-seismic electrical constraint term is constructed that includes rock physical statistical mapping relationships and formation interface geometric trends.

[0044] First, construct the statistical mapping constraint terms. Its discretization expression is: in, and The first The velocity and resistivity of each grid cell; An empirical regression function (such as a variant of the Gardner formula) pre-fitted based on the database described in step S4 is used to describe the statistical correlation between resistivity and wave velocity. A fixed coefficient vector that does not participate in the inversion iteration; These are local weighting coefficients based on rock physics confidence levels.

[0045] Secondly, construct interface trend constraints. The construction method of this item includes: (1) constructing tensor analysis on the joint inversion model output in step S3, calculating the gradient outer product matrix for each grid point and performing eigenvalue decomposition, and selecting the eigenvector corresponding to the largest eigenvalue as the prior stratum normal vector of that point. (2) Calculate the vertical gradient of physical parameters at the wellbore trajectory based on well logging data, and construct a reference gradient scalar field covering the entire area using spatial interpolation algorithms (such as Kriging interpolation). (3) Construct the distance-weighted constraint formula: in, The model parameters representing the current inversion ( or ); This represents the projection of the current model gradient onto the prior normal direction (i.e., the normal derivative). For the first Euclidean distance from each grid node to the nearest wellbore trajectory; The spatial distance weighting function is defined as follows: In the formula The normalization constant is The decay exponent, This is a stable term. The formula enforces the constraint that the model gradient conforms to logging characteristics near the wellbore, while constraining it to conform to the geometric trend of structural tensor extraction in regions far from the wellbore.

[0046] Finally, using the joint inversion model pair output in step S3 as the starting model, the above-mentioned well seismic electrical constraint term is introduced to construct the constrained second-order inversion objective function. Then, perform a constrained second inversion. The objective function is: In this function, and It remains a pure data fitting term; For cross gradient terms; and These are the prior constraints defined above. It should be noted that this objective function... The generic model roughness regularization term used in steps S1 and S2 is no longer included. and This is to avoid the general smoothing constraint blurring the fine stratigraphic boundaries introduced by the interface trend constraint term.

[0047] Through the above-mentioned secondary inversion, the final three-dimensional geological model of the salt cavern hydrogen storage cluster is output.

[0048] S6: Evaluation of hydrogen sealing performance based on the final three-dimensional geological model.

[0049] Based on the final constructed three-dimensional geological model, further evaluation of hydrogen sealing performance can be conducted. Combining the formation fluid samples collected in step S4 and the measured mechanical parameters, a three-dimensional geostress field model of the geological body is constructed. The hydrogen seepage control equation is introduced to simulate the hydrogen migration path under high-pressure hydrogen storage conditions, generating a hydrogen escape risk cloud map, thereby delineating the safe storage boundary.

[0050] S7: Four-dimensional dynamic geological monitoring during the operation of hydrogen storage facilities.

[0051] During the operation of the hydrogen storage facility, four-dimensional dynamic geological monitoring can be implemented. Based on the final geological model, differential inversion is performed on time-shifted electromagnetic and microseismic data collected at different time steps to calculate the rate of change of formation parameters over time. This allows for the identification of creep deformation areas in the salt cavity and the frontal diffusion range of hydrogen, ensuring the long-term safe operation of the storage facility.

[0052] In summary, the beneficial effects of the technical solution provided by this invention include: (1) It adopts a two-stage modeling process that combines "primary joint inversion" and "constrained secondary inversion". First, a joint inversion model pair with macroscopic structural consistency is constructed using the cross-gradient constraints of electromagnetic and seismic data. Then, using this as the starting model, well-seismic-electric constraint terms (including rock physical statistical mapping relationships and formation interface geometric trends) constructed using drilling data are introduced in the secondary inversion stage. This technical approach uses the results of primary inversion to provide a good initial structural framework for subsequent steps, avoiding the boundary distortion that may occur when directly using sparse well data for interpolation modeling. At the same time, the secondary inversion quantitatively corrects the model by introducing drilling measured data as strong prior information, thereby ensuring that the model conforms to the macroscopic geological structure, improving the consistency between the model's physical parameter values ​​and the measured data, and reducing the ambiguity of single geophysical inversion.

[0053] (2) The formation normal vector of the joint inversion model was extracted by constructing tensor analysis, and the normal gradient at the well trajectory was extended into a normal gradient trend field covering the entire area by combining spatial interpolation algorithm, thus constructing the interface trend constraint term. This technical feature extends the one-dimensional logging information that traditionally only exists in the well trajectory into a three-dimensional geometric trend constraint that can control the entire field. This allows the constructed geological model to maintain a layered morphology that conforms to the geological sedimentary laws or structural characteristics even in areas far from the well, thus improving the problem of unnatural model morphology that easily occurs in areas with complex geological structures using pure mathematical interpolation methods.

[0054] (3) By using discretized cross-gradient constraints, electromagnetic data, which is more sensitive to fluids and low-resistivity bodies, is jointly inverted with seismic data, which is more sensitive to the impedance interface of the formation waves. This technique forces the gradient vectors of the resistivity model and the velocity model to tend to be parallel or opposite during the iteration process, thereby achieving spatial consistency between the two at the geological structure boundary. This coupling mechanism uses the structural resolution capability of seismic data to compensate for the resolution deficiency caused by the volume effect of the electromagnetic method, and at the same time uses electromagnetic data to supplement the signal loss in the homogeneous zone or energy shielding zone inside the salt rock during seismic exploration, so that the final model can more clearly present the physical property interface between the salt rock and the surrounding rock.

[0055] The specific embodiments of the present invention described above do not constitute a limitation on the scope of protection of the present invention. Any other corresponding changes and modifications made in accordance with the technical concept of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies, characterized in that, Includes the following steps: S1: Organize historical geological data of the target area, conduct preliminary screening for database suitability, and delineate areas with concentrated salt resources; scan the delineated area using electromagnetic exploration technology, and perform inversion based on electromagnetic data to obtain a preliminary resistivity inversion model; S2: Deploy a 3D seismic detection system based on the electromagnetic exploration results, collect seismic wave data, and perform independent inversion to obtain a preliminary seismic wave velocity inversion model; S3: Perform grid resampling and matching, using the preliminary resistivity inversion model from step S1 and the preliminary seismic wave velocity inversion model from step S2 as initial guess models, and perform a primary joint inversion based on cross-gradient constraints on electromagnetic and seismic data to construct a joint inversion model pair that includes macroscopic structural consistency; 4: Based on the interpretation results of the joint inversion model pair, select the proposed reservoir area for drilling and sampling, and collect rock cores and formation fluid samples; conduct rock mechanics experiments, porosity and permeability tests, and physical property tests to construct a depth-rock mechanics parameter-resistivity-wave velocity multiphysics database and a depth-rock interface database; S5: Using the lithological data obtained from drilling as strong prior information, construct a well-seismic-electric constraint term that includes rock physical statistical mapping relationships and formation interface geometric trends; use the joint inversion model pair output in step S3 as the starting model for the secondary inversion, introduce the well-seismic-electric constraint term to construct a constrained secondary inversion objective function, perform constrained secondary inversion, and output the final three-dimensional geological model of the salt cavern hydrogen storage reservoir group.

2. The method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies as described in claim 1, characterized in that, In step S1, the electromagnetic data inversion employs a regularized Gauss-Newton inversion method to logarithmize the subsurface resistivity parameters, denoted as... Define the electromagnetic inversion objective function. Fitting terms from electromagnetic inversion data and electromagnetic inversion regularization term composition: The electromagnetic inversion data fitting term is defined as follows: In the formula For electromagnetic data weighting matrix, This is a vector of measured electromagnetic observation data. Here, we define the electromagnetic forward modeling function for calculating theoretical electromagnetic response data based on the resistivity model; the electromagnetic inversion regularization term is defined as follows: In the formula This is the discretization matrix of the three-dimensional spatial gradient operator; Regularization factor; model parameters The iterative update formula is: in, The electromagnetic forward modeling function Regarding model parameters The Jacobian matrix.

3. The method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies as described in claim 2, characterized in that, In step S2, the initial seismic wave velocity model is constructed using an optimization algorithm based on L-BFGS, and the seismic wave velocity parameters are logarithmically processed. Define the seismic inversion objective function. Fitting terms from seismic inversion data and earthquake inversion regularization term composition: Among them, the seismic inversion data fitting term is defined as In the formula Weighted matrix for earthquake data, This is a seismic forward modeling operator for calculating theoretical seismic wavefield data based on a velocity model. For seismic observation data; the seismic inversion regularization term is defined as follows: Here is the regularization factor for the velocity model; its iterative formula is: in: Step size factor It is an approximate Hessian inverse matrix. The gradient of the objective function.

4. The method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies as described in claim 3, characterized in that, In step S3, the specific steps of the joint inversion include: resampling and mapping the electromagnetic inversion grid to the seismic inversion grid to establish a unified inversion grid system; and constructing the joint inversion objective function. ： in, and These are the fitting terms for electromagnetic inversion data and the fitting terms for seismic inversion data, respectively. and These are the electromagnetic inversion regularization term and the seismic inversion regularization term, respectively; The discretized cross gradient constraint term is defined as follows: In the formula The total number of effective grid cells, and For the first Gradient vectors at each grid point; to These are the weighting coefficients.

5. The method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies as described in claim 1, characterized in that, In step S4, the criteria for selecting the proposed reservoir area are: based on the joint inversion model, geological interpretation is performed to infer that the salt layer burial depth is between 800 and 1500 m and the proportion of pure salt rock layer exceeds 80%; the test content includes porosity and permeability test and biochemical suitability evaluation.

6. The method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies as described in claim 4, characterized in that, In step S5, statistical mapping constraints are constructed using the lithological data obtained from drilling. and interface trend constraints The statistical mapping constraint term The discretized expression is: in, and The first The velocity and resistivity of each grid cell; The empirical regression function is pre-fitted based on the database described in step S4. The fixed coefficient vector that does not participate in the inversion iteration; the interface trend constraint term The construction method is as follows: (1) Perform tensor analysis on the joint inversion model output in step S3, calculate the gradient outer product matrix for each grid point and perform eigenvalue decomposition, and select the eigenvector corresponding to the largest eigenvalue as the formation normal vector of that point. (2) Calculate the normal gradient at the wellbore trajectory based on the logging data, and construct a normal gradient trend field covering the entire area using a spatial interpolation algorithm. subscript (3) Construct constraint formulas: ,in: For the set of inverted grid nodes; Represents model parameters At the node gradient vector at point With the normal vector of the strata at that location The dot product; These are the normalized weighting coefficients.

7. The method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies as described in claim 6, characterized in that, In step S5, the constrained quadratic inversion objective function for: in, and These are the fitting terms for electromagnetic inversion data and the fitting terms for seismic inversion data, respectively. For discretized cross-gradient constraint terms; For statistical mapping constraints; and These are the interface trend constraints for the resistivity model and the velocity model, respectively. Weighting coefficients to maintain data consistency and structural coupling. The weighting coefficients are used to control the strength of prior constraints.

8. The method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies as described in claim 1, characterized in that, The method for constructing the initial guessing model in step S3 further includes: using historical geological profiles around the target area to create a high-resolution geological model as a label; performing forward modeling and smoothing on the label to generate a low-resolution physical parameter model as input; constructing and training a structure recovery convolutional neural network; inputting the preliminary resistivity inversion model obtained in step S1 and the preliminary seismic wave velocity inversion model obtained in step S2 into the trained neural network to predict a physical parameter model containing fine structural features; and using the predicted model as the initial guessing model for the primary joint inversion in step S3.

9. The method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies as described in claim 1, characterized in that, It also includes a hydrogen sealing performance evaluation step based on the final three-dimensional geological model. The hydrogen sealing performance evaluation step includes: combining the formation fluid samples collected in step S4 to conduct a biochemical suitability evaluation, analyzing the content of sulfate-reducing bacteria and hydrogen consumption reaction parameters; and constructing a three-dimensional geostress field based on the final model to simulate the hydrogen migration path and generate a risk cloud map.

10. The method for constructing a geological model of a salt cavern hydrogen storage cluster based on the fusion of multiple technologies as described in claim 1, characterized in that, It also includes a four-dimensional dynamic geological monitoring step during the operation of the hydrogen storage facility. The four-dimensional dynamic geological monitoring step during the operation of the hydrogen storage facility includes differential inversion of time-shifted electromagnetic and microseismic data collected at different time steps based on the final geological model, calculating the rate of change of formation parameters, and identifying salt cavity creep and hydrogen diffusion range.

Images