Gas permeability monitoring method and system based on carbon dioxide geological storage

Abstract

The present invention discloses a gas permeability monitoring method and system based on carbon dioxide geological storage, which relates to the technical field of gas permeability monitoring, including obtaining geological data of a target stratum and a first constraint condition; perturbing the seismic wave frequency to generate a plurality of seismic monitoring control sets; constructing a three-dimensional geological model of the storage area and a wave velocity change simulation model, performing forward simulation, and obtaining wave velocity change response data under different frequency combinations; obtaining the wave velocity disturbance characteristic distribution caused by carbon dioxide injection under different seismic monitoring control sets, and weighted training a set evaluation model; constructing a second set according to the evaluation results of the set evaluation model and the corresponding seismic monitoring control set; fitting a multidimensional surface according to the second set, and screening the optimal seismic monitoring control set for seismic data acquisition and inversion, thereby solving the problem that traditional single-band inversion cannot dynamically match stratum structure and permeability characteristics.

Description

Technical Field

[0001] The present invention relates to the technical field of gas permeability monitoring, and more particularly, to a gas permeability monitoring method and system based on geological storage of carbon dioxide. Background Art

[0002] Geological carbon dioxide storage involves injecting captured carbon dioxide into deep underground formations (such as depleted oil and gas reservoirs, deep salt caverns, and aquifers) for long-term storage, preventing it from entering the atmosphere and mitigating its greenhouse gas effects. This is an important carbon capture and storage (CCS) technology.

[0003] Gas permeability monitoring refers to the use of sensors, data analysis, modeling algorithms or other technical means to obtain and analyze the permeability of underground formations to gas in real time or periodically, monitor the sealing effect, detect potential leaks, etc.

[0004] In the geological storage of carbon dioxide (CO2), inversion of seismic velocity changes is a key physical monitoring method, primarily used to assess changes in formation properties after CO2 injection, particularly identifying CO2 migration paths, diffusion range within the injection layer, and storage safety. The core principle is that the propagation velocity of seismic waves in formations is affected by factors such as lithology, porosity, fluid type, and saturation. When CO2 injection displaces the original fluid in the formation or alters pore pressure, this leads to changes in seismic velocity. By performing differential analysis on seismic data at different time points, subtle changes in seismic velocity can be inverted to infer the distribution and diffusion dynamics of CO2.

[0005] Specifically, this method is typically implemented using time-lapse seismic data. This involves collecting baseline seismic data before injection and multiple post-injection moments, extracting traveltime variations or waveform differences through consistency processing, and then reconstructing a model of formation wave velocity variations using an inversion algorithm. This process is highly dependent on the signal-to-noise ratio, acquisition parameter consistency, and frequency control strategy, with frequency control being particularly critical. Low frequencies help detect deep areas and large-scale changes, while high frequencies help distinguish details and anomaly boundaries. The combined use of these two improves the resolution and stability of the inversion.

[0006] Through the inversion of seismic wave velocity changes, not only can the migration process of carbon dioxide be tracked dynamically, but it can also help evaluate the integrity of the storage layer, the effectiveness of the cover rock and the potential leakage risk. Therefore, it plays an irreplaceable role in the long-term monitoring of CCS projects.

[0007] For example, the invention patent publication number CN119067255A discloses a method for predicting formation parameters for carbon dioxide geological storage. This method, which belongs to the field of oil and gas exploration and development, includes: obtaining first and second geological parameters of a target formation; obtaining first predicted formation parameters of the target formation based on the first geological parameters based on a first target prediction model for predicting the first predicted formation parameters, wherein the first predicted formation parameters include a predicted value for formation pressure, a predicted value for carbon dioxide saturation, and a predicted value for formation fluid density; and obtaining second predicted formation parameters of the target formation based on the first and second geological parameters based on a second target prediction model for predicting the second predicted formation parameters, wherein the second predicted formation parameters include a predicted value for mineral volume fraction and a predicted value for formation permeability. This application can improve the efficiency of predicting formation parameters during carbon dioxide geological storage.

[0008] For example, the invention patent announcement with announcement number CN118798425A discloses a method and device for predicting the storage capacity of carbon dioxide geological storage, including the following steps: preprocessing the collected carbon dioxide storage capacity data to obtain characteristic data; screening characteristic indicators from a preset indicator library through characteristic screening; wherein the preset indicator library includes several storage indicators that affect carbon dioxide geological storage; training a preset model based on the characteristic data and the several characteristic indicators to obtain a storage capacity prediction model; and predicting the storage capacity of carbon dioxide geological storage in the test area using the storage capacity prediction model. The present invention can improve the efficiency and accuracy of carbon dioxide geological storage storage capacity prediction.

[0009] The aforementioned technical solution presents at least the following technical issues: During the inversion of seismic velocity changes for CO2 geological storage, the frequency of the seismic waves plays a key role in monitoring accuracy and penetration depth. Setting the frequency too high improves inversion resolution and makes it easier to identify small-scale permeability changes, but also leads to severe signal attenuation, reduced penetration, and susceptibility to noise interference, resulting in unstable inversion results for deep targets. Setting the frequency too low, while providing good penetration and the ability to construct an overall velocity field, lacks spatial resolution, making it difficult to capture local velocity anomalies and prone to missing small leakage channels or early signs of infiltration.

[0010] Existing technologies mostly use a single frequency band for inversion, which lacks a dynamic matching mechanism between the depth changes of the stratum structure, the permeability scale characteristics and the signal characteristics, which can easily cause deviations in the inversion results or information missing, affecting the timeliness and accuracy of storage safety assessment and risk warning.

[0011] In view of the above problems, the present invention proposes a solution. Summary of the Invention

[0012] To overcome the aforementioned shortcomings of the prior art, embodiments of the present invention provide a system for monitoring gas permeability for CO2 geological storage. This system utilizes multi-frequency perturbations combined with various seismic sources and acquisition methods to construct a diverse seismic monitoring solution. This system utilizes forward modeling using three-dimensional geological and velocity simulations, combined with machine learning, to perform a weighted evaluation of the solutions and ultimately select the optimal frequency. This system addresses the inability of traditional single-band inversion to dynamically match stratum structure and permeability characteristics, significantly improving monitoring accuracy and the timeliness of storage safety assessments.

[0013] To achieve the above object, the present invention provides the following technical solutions:

[0014] A gas permeability monitoring method based on carbon dioxide geological storage includes the following steps: obtaining geological data of a target stratum and determining a first constraint condition required for inversion; perturbing the seismic wave frequency based on the first constraint condition, and generating a number of seismic monitoring control sets by randomly combining different source types and acquisition methods; constructing a three-dimensional geological model and a wave velocity variation simulation model of the storage area, and performing forward simulation based on each seismic monitoring control set to obtain wave velocity variation response data under different frequency combinations; obtaining the characteristic distribution of wave velocity disturbance caused by carbon dioxide injection under different seismic monitoring control sets, and weighting a set evaluation model based on the wave velocity variation response data; constructing a second set based on the evaluation results of the set evaluation model and the corresponding seismic monitoring control set; fitting a multidimensional surface based on the second set, and screening the optimal seismic monitoring control set for seismic data acquisition and inversion.

[0015] In a preferred embodiment, the geological data of the target formation is obtained and the first constraint condition required for inversion is determined, specifically: based on the core test results and the empirical model, the lithology type and its corresponding porosity and permeability characteristics are obtained; based on the core test fitting method, the lithology type and its corresponding porosity and permeability characteristics are converted into sensitivity factors of wave velocity disturbance, forming a lithology-disturbance response matrix, and quantitatively expressing the degree of wave velocity change caused by the penetration of carbon dioxide by different lithologies; according to the lithology-disturbance response matrix, the pressure gradient, injection rate and formation noise level in the injection history and predicted working conditions are obtained to set the first constraint condition of inversion.

[0016] In a preferred embodiment, the seismic wave frequency is disturbed based on the first constraint condition, and is randomly combined with different source types and acquisition methods to generate several seismic monitoring control sets, specifically: the first initial frequency is calculated based on geological data in combination with an empirical formula; the optimal frequency bandwidth range is obtained by combining the stratum scattering characteristics and signal-to-noise ratio analysis, and the stratum scattering characteristics and signal-to-noise ratio analysis include satisfying the maximum signal-to-noise ratio and covering the preset geological interface response frequency band; the first initial frequency is adjusted to capture the dynamic wave velocity changes caused by penetration, and the frequency point that matches the time constant of the penetration process is used as the second initial frequency; the second initial frequency is randomly perturbed while satisfying the first constraint condition to generate several frequencies to be screened; and several frequencies to be screened are randomly combined with different source types and acquisition methods to generate several seismic monitoring control sets.

[0017] In a preferred embodiment, the construction of a three-dimensional geological model of the storage area and a wave velocity variation simulation model is specifically as follows: based on the drilling, logging and seismic data of the target formation, a three-dimensional geological model of the storage area is constructed; the geological model is gridded; the pore pressure changes and saturation changes caused by carbon dioxide injection are mapped into physical responses of wave velocity changes; based on the finite difference method, the wave velocity changes are forward simulated to obtain the wave field responses under different combinations of seismic wave frequencies, source types and acquisition parameters, and the spatial distribution characteristics of the wave velocity disturbance are obtained; and through analysis of the simulation results, the correspondence between the wave velocity changes and the carbon dioxide permeation characteristics is extracted as the model output result.

[0018] In a preferred embodiment, forward simulation is performed based on each seismic monitoring and control set to obtain wave velocity change response data under different frequency combinations, specifically: forward simulation is performed on each seismic monitoring and control set based on the geological model and the wave velocity change simulation model, the characteristics of the simulation results are extracted and associated with the formation position and time to obtain wave velocity change response data.

[0019] In a preferred embodiment, the method of obtaining the characteristic distribution of wave velocity disturbance caused by carbon dioxide injection under different seismic monitoring control sets is specifically as follows: in the forward simulation results corresponding to each control set, the time series wave velocity disturbance images of the carbon dioxide injection area and its adjacent areas are identified, and the disturbance characteristic distribution is extracted; the disturbance characteristic distribution includes the disturbance morphological boundary of the injection impact area to form a spatial mask of the disturbance area; the volume change rate of the disturbance area at different times, the advancement speed and direction of the disturbance boundary; the wave velocity gradient distribution and uniformity index inside the disturbance area; the degree of matching between the main propagation path representing the disturbance front and the preset penetration direction; and the disturbance characteristic distribution is constructed as a disturbance characteristic vector.

[0020] In a preferred embodiment, a weighted training set evaluation model is combined with wave velocity change response data, specifically: a gradient boosting tree algorithm is used to construct a set evaluation model; the model uses a combination feature consisting of a disturbance feature vector and corresponding wave velocity change response data as input to construct a training sample; the training sample includes the disturbance feature vector and the wave velocity change response data; according to the inversion requirements of the target formation, a set evaluation index is set as a label value for supervised learning; the label value includes an effective disturbance recognition capability feature based on an assessment of the spatial matching degree between the disturbance distribution and the target storage layer; a time response accuracy feature based on an assessment of the first appearance time of the disturbance feature and the injection synchronization; a wave velocity disturbance structure clarity feature based on an assessment of the readability of the disturbance profile in the key area; based on the input and supervised learning labels, a gradient boosting tree model is used for regression training to output a set evaluation model.

[0021] In a preferred embodiment, the second set is constructed based on the evaluation results of the set evaluation model and the corresponding earthquake monitoring and control set, specifically; each second set includes an earthquake monitoring and control set and the output value of a corresponding set evaluation model.

[0022] In a preferred embodiment, the multidimensional surface is fitted according to the second set to screen the optimal seismic monitoring control set for seismic data acquisition and inversion, specifically: the second set is mapped to a multidimensional coordinate system, and several mapped points of the second set are fitted into a surface by a curve fitting method; the surface is differentiated, and the seismic monitoring control set corresponding to the most stable peak is analyzed to obtain the seismic monitoring control set for seismic data acquisition and inversion.

[0023] A system for monitoring gas permeability based on carbon dioxide geological storage includes a constraint module, a control set module, a geology module, a set evaluation module, and a screening module. The constraint module is used to obtain geological data of a target stratum and determine a first constraint condition required for inversion. The control set module is used to perturb the seismic wave frequency based on the first constraint condition and randomly combine different source types and acquisition methods to generate several seismic monitoring control sets. The geology module is used to construct a three-dimensional geological model and a wave velocity variation simulation model of the storage area, and perform forward simulation based on each seismic monitoring control set to obtain wave velocity variation response data under different frequency combinations. The set evaluation module is used to obtain the characteristic distribution of wave velocity disturbance caused by carbon dioxide injection under different seismic monitoring control sets, and to weight the wave velocity variation response data to train the set evaluation model. The screening module is used to construct a second set based on the evaluation results of the set evaluation model and the corresponding seismic monitoring control set. A multidimensional surface is fitted based on the second set to screen the optimal seismic monitoring control set for seismic data acquisition and inversion.

[0024] The technical effects and advantages of the gas permeability monitoring method and system based on carbon dioxide geological storage of the present invention are as follows:

[0025] 1. This invention generates a diverse set of seismic monitoring control sets by perturbing seismic wave frequencies based on geological constraints and combining them with random combinations of different source types and acquisition methods. This innovation transcends the limitations of traditional single-frequency monitoring, effectively covering the depth variations of strata structure and the characteristics of different permeability scales, achieving dynamic matching, and improving the ability to capture wave velocity perturbations and the accuracy of inversion, thereby enhancing the precision and reliability of CO2 storage gas permeability monitoring.

[0026] 2. This method constructs a detailed three-dimensional geological model and maps the changes in pore pressure and saturation caused by CO2 injection into the physical response of velocity changes. This method then uses the finite difference method to perform multi-parameter forward modeling. This method accurately reproduces the spatial distribution of velocity perturbations within the storage area, helps reveal the inherent connection between velocity perturbations and CO2 permeation, and enhances the physical authenticity of the simulation results and the fundamental reliability of the inversion model.

[0027] 3. This invention extracts detailed velocity perturbation characteristic distributions from forward simulation results, including multi-dimensional indicators such as perturbation region morphology, expansion velocity, and velocity gradient. Combined with velocity change response data, this method uses a gradient boosting tree algorithm to construct an ensemble evaluation model. This innovation enables intelligent weighted evaluation of seismic monitoring and control ensembles, significantly improving the timeliness and accuracy of inversion results, reducing the risk of subjective selection, and achieving automated and efficient monitoring solution optimization.

[0028] 4. This invention constructs a second set using the output of the ensemble evaluation model. Using a multidimensional surface fitting method, the discrete evaluation results are mapped into a continuous surface. Furthermore, the optimal seismic monitoring control set is selected by analyzing the surface derivatives to identify the most stable peak point. This technical approach effectively avoids local optimality traps, ensures the global optimality of the monitoring scheme in a multi-parameter space, enhances the stability and practicality of the acquisition and inversion scheme, and ensures the scientific nature and accuracy of the storage safety assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 Schematic diagram of the flow of the gas permeability monitoring method based on geological storage of carbon dioxide according to the present invention;

[0030] Figure 2 This is a schematic structural diagram of a gas permeability monitoring system based on geological storage of carbon dioxide according to the present invention. DETAILED DESCRIPTION

[0031] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0032] Example 1, Figure 1 The present invention provides a gas permeability monitoring method based on geological storage of carbon dioxide, comprising the following steps:

[0033] S1, obtaining geological data of the target stratum and determining the first constraint condition required for inversion.

[0034] In this embodiment, geological data of the target stratum is obtained and the first constraint condition required for inversion is determined, specifically:

[0035] Based on core test results and empirical models, obtain the lithology types and their corresponding porosity and permeability characteristics;

[0036] Based on the core test fitting method, the lithology and its corresponding porosity and permeability characteristics are converted into sensitivity factors to wave velocity disturbance, forming a lithology-disturbance response matrix to quantitatively express the degree of wave velocity change caused by carbon dioxide penetration in different lithologies.

[0037] The first constraint condition of the inversion is set based on the lithology-disturbance response matrix, combined with the injection history and the pressure gradient, injection rate and formation noise level in the predicted working conditions.

[0038] It should be noted that core testing is one of the primary means of obtaining the true physical properties of underground formations. By testing parameters such as permeability, porosity, density, and wave velocity on collected core samples, we can obtain data on the porosity and permeability characteristics of different lithologies under experimental conditions. Empirical models such as the Kozeny-Carman equation and the Wyllie time-averaged equation are used to estimate and supplement porosity and permeability characteristics based on established empirical rules when test data is insufficient or inaccurate, ensuring the formation of a complete and reliable database of formation physical properties.

[0039] It should be noted that lithology types generally include sandstone, shale, limestone, siltstone, etc. Each lithology has different porosity and permeability due to different mineral composition, particle structure and degree of cementation. Porosity reflects the size of the rock storage space, while permeability affects the The ability of rocks to flow in pores. The response intensity of pore pressure and saturation changes caused by injection is the basic physical property that affects wave velocity disturbance.

[0040] Core testing fitting involves combining actual core test data with seismic interpretation data and well logging curves to construct a mapping relationship between lithology, porosity, and velocity perturbations. This method typically employs multivariate regression or machine learning to identify correlations between lithologic characteristics (such as shale content and grain size distribution) and their corresponding velocity variations, thereby establishing parameter sensitivity models that can be used in seismic wave simulations.

[0041] The wave velocity perturbation sensitivity factor is used to measure the sensitivity of a certain rock type to the wave velocity perturbation. The intensity of the wave velocity response after injection is calculated based on core experimental data, laboratory injection test results and numerical simulation results. Its value is expressed as the rate of change of wave velocity caused by unit pressure change or saturation change, which is used for accurate modeling. Elastic wave response under the influence of injection.

[0042] The lithology-disturbance response matrix is a two-dimensional array, with rows representing different lithologies and columns representing disturbance conditions (such as pressure increment, saturation change). The matrix cells are the velocity perturbation sensitivity factors under the corresponding conditions. This matrix is the core input for forward modeling and inversion constraints, and can be used to distinguish the effects of different lithologies on the The response differences of the infiltration process can improve the simulation accuracy.

[0043] Injection history includes Injection start time, cumulative injection volume, injection rate curve, and well layout are important indicators for establishing the stratigraphic evolution context. Predictive conditions, based on geological storage planning, simulate and predict injection conditions (such as rate and pressure) over a period of time to assess the long-term response characteristics of the storage area.

[0044] The first constraint combines the lithologic response matrix, injection parameters, and noise level to form the control boundary conditions for subsequent forward and inversion modeling. It defines a reasonable frequency perturbation range, sensitive formation areas, and the expected range of velocity response, providing a technical foundation for designing efficient seismic monitoring control ensembles.

[0045] The construction of the above-mentioned first constraint condition realizes the physical response mapping between geological properties and wave velocity disturbances by constructing a lithology-disturbance response matrix and setting the first constraint condition of inversion. Therefore, it is possible to accurately evaluate the sensitivity of different lithologies to the wave velocity changes caused by the carbon dioxide penetration process without relying on a large amount of field monitoring data, effectively improving the pertinence and scientific nature of subsequent frequency disturbance simulation and seismic monitoring parameter design, and thus enhancing the inversion model's ability to identify the permeability characteristics in complex storage formations, with significant predictive accuracy and engineering practical value.

[0046] S2, based on the first constraint, perturbs the seismic wave frequency and randomly combines different source types and acquisition methods to generate several seismic monitoring control sets.

[0047] In this embodiment, based on the first constraint, the seismic wave frequency is disturbed, and different source types and acquisition methods are randomly combined to generate several seismic monitoring control sets, specifically:

[0048] Calculate the first initial frequency based on geological data combined with empirical formula;

[0049] Combining formation scattering characteristics with signal-to-noise ratio analysis to obtain an optimal frequency bandwidth range, wherein the formation scattering characteristics and signal-to-noise ratio analysis include satisfying a maximum signal-to-noise ratio and covering a preset geological interface response frequency band;

[0050] Adjust the first initial frequency to capture the dynamic wave velocity changes caused by penetration, and use the frequency point that matches the time constant of the penetration process as the second initial frequency;

[0051] Randomly perturb the second initial frequency while satisfying the first constraint condition to generate a number of frequencies to be screened;

[0052] Several frequencies to be screened are randomly combined with different earthquake source types and acquisition methods to generate several earthquake monitoring control sets.

[0053] It should be noted that geological data, including well logging data, lithologic profiles, interlayer thickness, and velocity curves, reflect the physical structure and dielectric properties of the formation. By combining existing empirical formulas (such as the frequency-velocity relationship derived from formation shear modulus and density), we can preliminarily estimate the initial frequency that effectively penetrates the target formation and generates a significant reflection signal. This frequency, known as the first initial frequency, provides a baseline for subsequent frequency perturbations.

[0054] Stratum scattering refers to the dispersion of seismic wave propagation energy caused by geological interfaces, heterogeneous structures, and fractures. High-frequency signals are more affected by scattering and absorption, while low-frequency signals have limited resolution. Signal-to-noise ratio analysis combines background noise, instrument response, and geological interface reflection characteristics in actual exploration environments to identify the frequency bandwidth that maximizes the signal-to-noise ratio while maintaining sufficient resolution, thereby determining the scope of subsequent disturbances.

[0055] Permeability in storage formations typically exhibits a dynamic timescale, reflected in the seismic response as dynamic changes within a specific frequency range. By introducing a time constant (such as Darcy flow or capillary front diffusion rate), the frequency point most sensitive to this dynamic process is identified as the second initial frequency, allowing for more accurate capture of the velocity perturbations caused by permeability.

[0056] Disturbance is introduced on the basis of the second initial frequency, and a number of frequencies to be screened are generated using mathematical models such as normal distribution or uniform distribution. This simulates the impact of different settings on formation response in actual seismic monitoring and enhances the system's ability to identify sensitive intervals of frequency changes.

[0057] Source types such as explosive, hammer, or vibroseis, and acquisition methods including multi-component detection, linear array, and planar array are cross-combined with the frequencies to be screened to form multiple seismic monitoring control sets. Each set represents a specific monitoring strategy, independently executed in forward modeling to identify the parameter combination that optimally responds to the velocity perturbations caused by infiltration.

[0058] This application can effectively capture the initial frequency by accurately determining the initial frequency based on geological data and empirical formulas, and optimizing the frequency bandwidth by combining the scattering characteristics of the formation and the signal-to-noise ratio analysis. Dynamic velocity variations during the permeation process. Furthermore, through random combinations of frequency perturbations and various source types and acquisition methods, a diverse seismic monitoring control set was constructed, enhancing the adaptability and robustness of the monitoring scheme. This approach not only improves the sensitivity and resolution of seismic data to velocity perturbations in the storage area but also provides a scientifically sound parameter basis for subsequent forward simulations and inversion analyses, significantly improving the accuracy and reliability of gas permeability monitoring.

[0059] S3: Construct a three-dimensional geological model of the storage area and a wave velocity change simulation model, and perform forward simulation based on each seismic monitoring control set to obtain wave velocity change response data under different frequency combinations.

[0060] In this embodiment, a three-dimensional geological model of the storage area and a wave velocity variation simulation model are constructed, specifically:

[0061] Construct a 3D geological model of the storage area based on drilling, logging, and seismic data of the target formation;

[0062] Gridding the geological model;

[0063] Mapping the pore pressure and saturation changes caused by CO2 injection into physical responses of wave velocity changes;

[0064] Based on the finite difference method, the wave velocity changes are forward simulated to obtain the wave field response under different combinations of seismic wave frequencies, source types and acquisition parameters, and the spatial distribution characteristics of the wave velocity disturbance are obtained;

[0065] Through the analysis of simulation results, the corresponding relationship between wave velocity changes and carbon dioxide penetration characteristics is extracted as the model output result.

[0066] Furthermore, forward simulation is performed based on each earthquake monitoring control set to obtain the wave velocity change response data under different frequency combinations, specifically:

[0067] For each seismic monitoring control set, forward simulation is performed based on the geological model and wave velocity variation simulation model. The characteristics of the simulation results are extracted and associated with the formation position and time to obtain wave velocity variation response data.

[0068] It should be noted that the finite element or finite difference method is used to divide the three-dimensional geological model into regular or irregular grid cells to form a computational grid. Meshing must balance computational accuracy and efficiency to ensure that the spatial variations of the geological body can be accurately reflected while supporting the discrete calculation of the wave field in the numerical simulation.

[0069] Mapping changes in pore pressure and saturation caused by CO2 injection into the physical response of wave velocity changes: Based on fluid dynamics and rock physics theory, a physical relationship between pore pressure and rock elastic parameters (such as compressional wave velocity (Vp) and shear wave velocity (Vs)) is established. This allows the increase in pore pressure and changes in saturation caused by CO2 injection to be converted into changes in local wave velocity. This mapping process, based on poroelasticity theory, the Gassmann equation, or similar rock physics models, quantitatively expresses the impact of permeability dynamics within the storage area on wave velocity.

[0070] By analyzing the simulation results, the corresponding relationship between wave velocity changes and CO2 permeation characteristics was extracted as the model output. The wavefield response data obtained from the forward simulation was analyzed to identify the wave velocity perturbation patterns and their spatiotemporal distribution under different seismic parameter combinations. This relationship was then established between wave velocity changes and CO2 permeation processes (such as permeation rate, permeation path, and permeation range). This relationship provided a key physical basis and quantitative reference for subsequent wave velocity change inversion and permeability monitoring.

[0071] For each seismic monitoring control set generated in step S2 (including frequency perturbation, source type, and acquisition method combinations), forward simulations are performed based on the 3D geological model and velocity variation simulation model. By extracting key wavefield features from each simulation result (such as amplitude variations, spectral variations, and velocity perturbation intensity in time and space) and correlating them with specific stratigraphic locations and time points, a comprehensive and dynamic velocity variation response dataset is generated, laying the data foundation for subsequent evaluation model training and inversion analysis.

[0072] It should be noted that existing CO2 geological storage gas permeability monitoring technologies typically use a single frequency band for seismic data acquisition and inversion, ignoring the differences in wave velocity response of the stratum structure at different depths and the multi-scale characteristics of the permeability process. Single-frequency band inversion lacks a coupling and matching mechanism between stratum depth variations, permeability dynamics, and signal characteristics. As a result, the collected seismic signals may not fully reflect the complex permeability behavior within the storage area, leading to biased inversion results or information omissions. This limitation directly affects the timeliness and accuracy of CO2 storage safety assessments and risk warnings, increasing the potential risks of geological storage operations.

[0073] This step achieves comprehensive response capture of the stratigraphic structure at different depths and multi-scale infiltration processes in the storage area through forward simulation based on multi-frequency combinations and dynamic velocity change simulation. The combination of multiple frequencies, source types, and acquisition methods can more effectively extract the spatiotemporal distribution characteristics of velocity disturbances, significantly improving the accuracy and stability of velocity change inversion. This method promotes the dynamic matching of infiltration dynamic characteristics with seismic signals, improves the monitoring system's ability to identify carbon dioxide migration behavior, enhances the reliability of storage safety assessments and the timeliness of risk warnings, and contributes to more scientific and accurate geological storage management.

[0074] S4, obtain the characteristic distribution of velocity disturbance caused by CO2 injection under different seismic monitoring control sets, and use the velocity change response data to weight the training set evaluation model.

[0075] In this embodiment, the characteristic distribution of wave velocity disturbance caused by carbon dioxide injection under different seismic monitoring control sets is obtained, specifically:

[0076] In the forward simulation results corresponding to each control set, the time series wave velocity disturbance images of the CO2 injection area and its adjacent areas are identified, and the characteristic distribution of the disturbance is extracted;

[0077] The disturbance characteristic distribution includes the disturbance morphological boundary of the injection impact area, forming a disturbance area spatial mask;

[0078] The volume change rate of the disturbance area at different times, the speed and direction of the disturbance boundary advancement;

[0079] Wave velocity gradient distribution and uniformity index inside the disturbance area;

[0080] Represents the degree of matching between the main propagation path of the disturbance front and the preset penetration direction;

[0081] Construct the perturbed feature distribution as a perturbed feature vector.

[0082] Furthermore, the weighted training set evaluation model is combined with the wave velocity change response data, specifically:

[0083] The gradient boosting tree algorithm is used to construct the set evaluation model;

[0084] The model uses the combined features of the disturbance feature vector and the corresponding wave velocity change response data as input to construct training samples;

[0085] The training samples include disturbance feature vectors and wave velocity change response data;

[0086] According to the inversion requirements of the target formation, the set evaluation index is set as the label value of supervised learning;

[0087] The tag value includes an effective disturbance identification capability feature based on an evaluation of the spatial matching degree between the disturbance distribution and the target storage layer;

[0088] Time response accuracy characteristics based on the first appearance time of the disturbance feature and the injection synchronization evaluation;

[0089] The clarity of the wave velocity perturbation structure is assessed based on the interpretability of the perturbation profile in key areas;

[0090] Based on the input and supervised learning labels, the gradient boosting tree model is used for regression training and the output is a set evaluation model.

[0091] It should be noted that the recognition of time series wave velocity disturbance images

[0092] Based on the forward simulation results corresponding to each seismic monitoring and control set, the team utilized seismic wavefield data with high temporal and spatial resolution to identify the evolution of velocity perturbations within the injection zone and its adjacent areas. By extracting velocity variation images at consecutive time points, they generated a time series of velocity perturbation images, reflecting the dynamic response of the formation to CO2 injection.

[0093] Formation of disturbance morphological boundaries and spatial masks: Image processing and edge detection algorithms (such as the Canny operator and region growing) are used to extract the disturbance boundaries of the injection impact area and form a spatial mask of the disturbance area. The spatial mask is used to define the geographic scope of the wave velocity disturbance, facilitating subsequent quantification of disturbance characteristics and spatial matching analysis.

[0094] The volume change rate of the disturbed region and the velocity of its boundary advance: The spatial volume of the disturbed region at different times is calculated and the rate of change of the disturbed volume is analyzed to reveal the dynamic expansion trend of the CO2 penetration effect. Furthermore, by tracking the movement of the disturbance boundary over time, the velocity and direction of the boundary advance are calculated to characterize the propagation characteristics of the penetration wavefront.

[0095] Within the disturbance area, the spatial gradient of wave velocity is calculated to reflect the local intensity and directionality of the wave velocity change. Combined with statistical analysis (such as variance and entropy), the uniformity of the wave velocity disturbance is evaluated to characterize the homogeneous or heterogeneous characteristics of the infiltration process.

[0096] Based on the prior injection scheme and geological structure, the preset main infiltration direction is determined. The main propagation path of the disturbance front is calculated through path analysis methods (such as the shortest path algorithm or streamline tracing), and the matching degree is evaluated with the preset infiltration direction to verify the rationality of the wave velocity disturbance and the conformity of the infiltration process.

[0097] The above-mentioned quantitative features, such as disturbance morphological boundary, volume change rate, boundary velocity, wave velocity gradient, uniformity index and propagation path matching, are combined into a disturbance feature vector, which serves as the input data of the subsequent machine learning model to fully describe the wave velocity disturbance information caused by carbon dioxide injection.

[0098] It should be noted that the gradient boosting tree algorithm is used to train the ensemble evaluation model. GBT is an ensemble learning method that gradually reduces prediction errors through multiple rounds of iteration. It has good nonlinear modeling capabilities and robustness, and is suitable for processing complex multidimensional earthquake monitoring data and disturbance characteristics.

[0099] The disturbance eigenvectors are fused with the corresponding velocity response data to form a combined feature input. The velocity response data provides supplementary information on spatiotemporal wavefield characteristics, enhancing the model's ability to learn multi-dimensional formation responses.

[0100] Based on the inversion requirements of the target formation, appropriate ensemble evaluation metrics are designed as labels (target variables) for model training, guiding the model to learn the strengths and weaknesses of different seismic monitoring control ensembles. These labels cover three aspects: effective disturbance identification, temporal response accuracy, and clarity of velocity disturbance structure. This ensures that the comprehensive evaluation metrics output by the model are highly representative and practical.

[0101] Effective disturbance identification capability characteristics: Based on the degree of spatial matching between the disturbance distribution and the target storage layer, the ensemble's ability to capture velocity disturbances in key penetration areas is evaluated.

[0102] Time response accuracy characteristics: The synchronization between the first appearance time of the disturbance feature and the actual injection time is used as an indicator to reflect the real-time response capability of the monitoring system to the infiltration process.

[0103] Velocity disturbance structure clarity characteristics: evaluate the readability of the disturbance contour in key areas and reflect the spatial resolution and signal-to-noise ratio of the seismic velocity disturbance image.

[0104] Using labeled training samples, a gradient boosting tree model is used for regression training, gradually optimizing model parameters to enable the model to accurately predict the set evaluation value corresponding to the input features. Ultimately, the model outputs a scalar evaluation value that is used to comprehensively evaluate the quality of each earthquake monitoring and control set and assist in selecting the optimal acquisition plan.

[0105] This example constructs a perturbation feature vector and combines it with the combined features of velocity response data, then uses a gradient boosting tree algorithm to establish a set evaluation model. This method enables quantitative evaluation and intelligent screening of different seismic monitoring control sets. This method effectively captures the spatial dynamic characteristics and multidimensional velocity response of the infiltration process, improving the scientific nature and accuracy of monitoring schemes and providing solid data support and decision-making basis for ensuring the safety of geological carbon dioxide storage.

[0106] In the existing technology, seismic inversion for geological storage of carbon dioxide mostly uses a single frequency band to monitor and invert wave velocity changes, without fully considering the complex changes in the stratum structure in the depth direction and the multi-scale penetration characteristics during the carbon dioxide penetration process. This single frequency band inversion method lacks a dynamic matching mechanism for the different response characteristics of the stratum at different depths and the frequency characteristics of the seismic signal, which makes the inversion results prone to deviations and cannot fully reflect the spatial and temporal dynamic changes of the penetration process. In addition, ignoring frequency diversity may also result in the loss of key wave velocity disturbance information, reducing the assessment accuracy of the safety of the storage area and the timeliness of risk warnings. For this reason, there is an urgent need to develop a comprehensive inversion method that combines multi-frequency combinations and multi-parameter seismic monitoring control sets to improve the capture capability and recognition accuracy of the dynamic response of the stratum caused by carbon dioxide injection.

[0107] This step constructs a seismic monitoring control set with a random combination of multiple frequencies and different source types and acquisition methods, and conducts forward simulation based on a three-dimensional geological model and a wave velocity change simulation model to systematically obtain wave velocity change response data and wave velocity disturbance characteristic distribution under multi-frequency and multi-parameter conditions. This method realizes the dynamic matching of the structure and permeability scale characteristics of the stratum at different depths, significantly improving the spatial resolution and time response accuracy of the seismic inversion results. It effectively avoids the information omission and error accumulation caused by single-band inversion, and enhances the sensitivity and recognition ability of the carbon dioxide penetration process. This improves the accuracy of the safety assessment of the geological storage area and the timeliness of the risk warning, and provides more reliable technical support for the scientific decision-making of the storage monitoring system.

[0108] S5, constructing a second set according to the evaluation results of the set evaluation model and the corresponding earthquake monitoring control set.

[0109] Each second set includes an earthquake monitoring control set and an output value of a corresponding set evaluation model.

[0110] S6, fitting a multidimensional surface according to the second set, and screening the optimal seismic monitoring control set for seismic data acquisition and inversion.

[0111] In this embodiment, a multidimensional surface is fitted based on the second set to screen the optimal seismic monitoring control set for seismic data acquisition and inversion, specifically:

[0112] Mapping the second set to a multidimensional coordinate system, and fitting a plurality of mapped points of the second set into a curved surface by a curve fitting method;

[0113] The surface is differentiated and the seismic monitoring control set corresponding to the most stable peak is obtained for seismic data acquisition and inversion.

[0114] It should be noted that by calculating the partial derivatives (i.e., the gradients in each dimension) of the fitted multidimensional surface, the rate of change information for each point on the surface can be obtained. A stationary point is a point where the gradient approaches zero, indicating that the surface changes at this point tends to be gentle, with no significant upward or downward trends. Furthermore, by calculating the second-order derivative (i.e., the Hessian matrix) at this point, its properties are determined to ensure that it is a local maximum (i.e., the surface is concave in all directions at this point), thus confirming that it is a stable and optimal peak point. The seismic monitoring control set corresponding to this point is then selected as the optimal solution, ensuring the optimal and stable overall performance of the evaluation indicators, reducing uncertainty caused by indicator fluctuations, and improving the effectiveness and accuracy of seismic data acquisition and inversion.

[0115] Example 2, Figure 2 The present invention provides a gas permeability monitoring system based on carbon dioxide geological storage, including a constraint module, a control set module, a geological module, a set evaluation module, and a screening module;

[0116] A constraint module is used to obtain geological data of the target stratum and determine the first constraint condition required for inversion;

[0117] A control set module is used to perturb the seismic wave frequency based on the first constraint condition and randomly combine different source types and acquisition methods to generate several seismic monitoring control sets;

[0118] The geological module is used to construct a three-dimensional geological model of the storage area and a wave velocity variation simulation model, and to perform forward simulation based on each seismic monitoring control set to obtain wave velocity variation response data under different frequency combinations;

[0119] The ensemble evaluation module is used to obtain the characteristic distribution of velocity disturbance caused by CO2 injection under different seismic monitoring control sets, and to train the ensemble evaluation model based on the weighted velocity change response data;

[0120] The screening module is used to construct a second set based on the evaluation results of the set evaluation model and the corresponding seismic monitoring control set; fit a multidimensional surface based on the second set, and screen the optimal seismic monitoring control set for seismic data acquisition and inversion.

Claims

1. A gas permeability monitoring method based on geological storage of carbon dioxide, characterized in that: The steps include: Obtain geological data of the target stratum and determine the first constraint condition required for inversion; Based on the first constraint, the seismic wave frequency is disturbed and randomly combined with different source types and acquisition methods to generate several seismic monitoring control sets; Construct a three-dimensional geological model of the storage area and a wave velocity variation simulation model, and perform forward simulation based on each seismic monitoring control set to obtain wave velocity variation response data under different frequency combinations; Obtain the characteristic distribution of velocity disturbance caused by CO2 injection under different seismic monitoring control sets, and use the velocity change response data to weight the training set evaluation model; constructing a second set according to the evaluation results of the set evaluation model and the corresponding earthquake monitoring control set; A multidimensional surface is fitted according to the second set, and the optimal seismic monitoring control set is screened for seismic data acquisition and inversion.

2. The gas permeability monitoring method based on carbon dioxide geological storage according to claim 1, characterized in that: The geological data of the target stratum is obtained and the first constraint condition required for inversion is determined, specifically: Based on core test results and empirical models, obtain the lithology types and their corresponding porosity and permeability characteristics; Based on the core test fitting method, the lithology and its corresponding porosity and permeability characteristics are converted into sensitivity factors to wave velocity disturbance, forming a lithology-disturbance response matrix to quantitatively express the degree of wave velocity change caused by carbon dioxide penetration in different lithologies. The first constraint condition of the inversion is set based on the lithology-disturbance response matrix, combined with the injection history and the pressure gradient, injection rate and formation noise level in the predicted working conditions.

3. The gas permeability monitoring method based on carbon dioxide geological storage according to claim 2, characterized in that: Based on the first constraint, the seismic wave frequency is disturbed and randomly combined with different source types and acquisition methods to generate several seismic monitoring control sets, specifically: Calculate the first initial frequency based on geological data combined with empirical formula; Combining formation scattering characteristics with signal-to-noise ratio analysis to obtain an optimal frequency bandwidth range, wherein the formation scattering characteristics and signal-to-noise ratio analysis include satisfying a maximum signal-to-noise ratio and covering a preset geological interface response frequency band; Adjust the first initial frequency to capture the dynamic wave velocity changes caused by penetration, and use the frequency point that matches the time constant of the penetration process as the second initial frequency; Randomly perturb the second initial frequency while satisfying the first constraint condition to generate a number of frequencies to be screened; Several frequencies to be screened are randomly combined with different earthquake source types and acquisition methods to generate several earthquake monitoring control sets.

4. The gas permeability monitoring method based on carbon dioxide geological storage according to claim 3, characterized in that: The construction of the three-dimensional geological model of the storage area and the wave velocity change simulation model is specifically as follows: Construct a 3D geological model of the storage area based on drilling, logging, and seismic data of the target formation; Gridding the geological model; Mapping the pore pressure and saturation changes caused by CO2 injection into physical responses of wave velocity changes; Based on the finite difference method, the wave velocity changes are forward simulated to obtain the wave field response under different combinations of seismic wave frequencies, source types and acquisition parameters, and the spatial distribution characteristics of the wave velocity disturbance are obtained; Through the analysis of simulation results, the corresponding relationship between wave velocity changes and carbon dioxide penetration characteristics is extracted as the model output result.

5. The gas permeability monitoring method based on carbon dioxide geological storage according to claim 4, characterized in that: The forward simulation is performed based on each earthquake monitoring control set to obtain the wave velocity change response data under different frequency combinations, specifically: For each seismic monitoring control set, forward simulation is performed based on the geological model and wave velocity variation simulation model. The characteristics of the simulation results are extracted and associated with the formation position and time to obtain wave velocity variation response data.

6. The gas permeability monitoring method based on carbon dioxide geological storage according to claim 5, characterized in that: The method of obtaining the characteristic distribution of wave velocity disturbance caused by carbon dioxide injection under different seismic monitoring control sets is specifically as follows: In the forward simulation results corresponding to each control set, the time series wave velocity disturbance images of the CO2 injection area and its adjacent areas are identified, and the characteristic distribution of the disturbance is extracted; The disturbance characteristic distribution includes the disturbance morphological boundary of the injection impact area, forming a disturbance area spatial mask; The volume change rate of the disturbance area at different times, the speed and direction of the disturbance boundary advancement; Wave velocity gradient distribution and uniformity index inside the disturbance area; Represents the degree of matching between the main propagation path of the disturbance front and the preset penetration direction; Construct the perturbed feature distribution as a perturbed feature vector.

7. The gas permeability monitoring method based on carbon dioxide geological storage according to claim 6, characterized in that: Combined with the wave velocity change response data, the weighted training set evaluation model is as follows: The gradient boosting tree algorithm is used to construct the set evaluation model; The model uses the combined features of the disturbance feature vector and the corresponding wave velocity change response data as input to construct training samples; The training samples include disturbance feature vectors and wave velocity change response data; According to the inversion requirements of the target formation, the set evaluation index is set as the label value of supervised learning; The tag value includes an effective disturbance identification capability feature based on an evaluation of the spatial matching degree between the disturbance distribution and the target storage layer; Time response accuracy characteristics based on the first appearance time of the disturbance feature and the injection synchronization evaluation; The clarity of the wave velocity perturbation structure is assessed based on the interpretability of the perturbation profile in key areas; Based on the input and supervised learning labels, the gradient boosting tree model is used for regression training and the output is a set evaluation model.

8. The gas permeability monitoring method based on carbon dioxide geological storage according to claim 7, characterized in that: The second set is constructed according to the evaluation results of the set evaluation model and the corresponding earthquake monitoring control set, specifically: Each second set includes an earthquake monitoring control set and an output value of a corresponding set evaluation model.

9. The gas permeability monitoring method based on carbon dioxide geological storage according to claim 8, characterized in that: The multidimensional surface is fitted according to the second set to select the optimal seismic monitoring control set for seismic data acquisition and inversion, specifically: Mapping the second set to a multidimensional coordinate system, and fitting a plurality of mapped points of the second set into a curved surface by a curve fitting method; The surface is differentiated and the seismic monitoring control set corresponding to the most stable peak is obtained for seismic data acquisition and inversion.

10. A system using the gas permeability monitoring method based on carbon dioxide geological storage according to any one of claims 1 to 9, characterized in that: It includes constraint module, control set module, geological module, set evaluation module and screening module; A constraint module is used to obtain geological data of the target stratum and determine the first constraint condition required for inversion; A control set module is used to perturb the seismic wave frequency based on the first constraint condition and randomly combine different source types and acquisition methods to generate several seismic monitoring control sets; The geological module is used to construct a three-dimensional geological model of the storage area and a wave velocity variation simulation model, and to perform forward simulation based on each seismic monitoring control set to obtain wave velocity variation response data under different frequency combinations; The ensemble evaluation module is used to obtain the characteristic distribution of velocity disturbance caused by CO2 injection under different seismic monitoring control sets, and to train the ensemble evaluation model based on the weighted velocity change response data; The screening module is used to construct a second set based on the evaluation results of the set evaluation model and the corresponding seismic monitoring control set; fit a multidimensional surface based on the second set, and screen the optimal seismic monitoring control set for seismic data acquisition and inversion.

Images