Geological horizon automatic identification and modeling method and system

Abstract

The invention belongs to the technical field of geological exploration, and particularly relates to a geological horizon automatic identification and modeling method and system. The system comprises a seismic data preprocessing module, a horizon initial identification module, a fault sensing and adaptive correction module, a three-dimensional horizon curved surface construction module and a quality control and output module. The end-to-end automatic processing from original seismic data to the three-dimensional horizon model is realized, the dependence on external measured data is remarkably reduced, and the method is particularly suitable for an initial exploration stage or a well-free region.

Description

Technical Field

[0001] This invention belongs to the field of geological exploration technology, specifically relating to a method and system for automatic identification and modeling of geological strata. Background Technology

[0002] With the continuous advancement of geological exploration and modeling technologies, the automatic identification and 3D modeling of geological strata are playing an increasingly important role in key areas such as oil and gas resource development, marine engineering planning, and geological disaster early warning. As a core information carrier reflecting the distribution and evolutionary history of underground rock strata, the accurate reconstruction of geological strata relies on in-depth analysis of seismic data and effective fusion of multi-source information. Currently, geological modeling is gradually evolving from manual interpretation to intelligent and automated methods. However, achieving highly complete and robust automatic extraction of strata and continuous 3D modeling in the absence of prior drilling data remains a core challenge in this field.

[0003] The automatic geological horizon identification and modeling technology aims to extract geologically significant phase axes directly from the original seismic profile using algorithms and construct a three-dimensional horizon surface that conforms to geological laws. This process must consider the lateral continuity and vertical correspondence of the horizons, as well as adaptability to complex structures (such as faults and folds). An ideal technical solution should be able to complete the entire process from seismic data input to structured horizon model output end-to-end, without relying on external measured data or extensive manual intervention.

[0004] Existing technologies still have significant limitations in this direction. On the one hand, while some methods introduce image recognition or machine learning techniques for initial stratigraphic extraction, they heavily rely on repeated corrections using measured data such as well core sampling, resulting in limited application scope, high costs, and difficulty in applying them to the early stages of exploration or areas without wells. On the other hand, another type of technology focuses only on the post-processing optimization of existing stratigraphic data, such as removing outliers through fault distance calculations, but it cannot solve the problem of missing or fractured initial stratigraphic identification, and its effectiveness is highly dependent on the accuracy of the fault model itself. More importantly, existing solutions generally fail to organically integrate stratigraphic identification, fault sensing, and 3D modeling, leading to stratigraphic misconnections, interruptions, or geometric distortions in complex structural areas, making it difficult to meet the urgent needs of modern geological exploration for fully automated, high-precision, and highly adaptable modeling tools. Therefore, a new technical approach is urgently needed that can start from raw seismic data, embed fault adaptive mechanisms, and achieve end-to-end 3D modeling. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for automatic identification and modeling of geological strata, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: An automatic geological stratum identification and modeling system includes the following components: a seismic data preprocessing module for receiving raw seismic data and performing data standardization and signal-to-noise ratio enhancement operations; a stratum initial identification module connected to the seismic data preprocessing module for extracting a set of candidate stratum points from the preprocessed seismic data; a fault sensing and adaptive correction module connected to the stratum initial identification module for identifying fault structures and performing geometric correction on the candidate stratum points under fault influence; a three-dimensional stratum surface construction module connected to the fault sensing and adaptive correction module for generating a continuous three-dimensional stratum surface model based on the corrected stratum point set; and a quality control and output module connected to the three-dimensional stratum surface construction module for verifying the integrity of the generated stratum model and outputting a final structured geological model. This invention also includes a method for automatic identification and modeling of geological strata, the specific steps of which are as follows: Step S110: Preprocess the raw seismic data, including data standardization and signal-to-noise ratio enhancement operations; Step S120: Extract the candidate set of horizon points from the preprocessed seismic data. Step S130: Identify fault structures and perform geometric correction on candidate stratigraphic points under the influence of faults; Step S140: Generate a continuous three-dimensional stratigraphic surface model based on the corrected stratigraphic point set; Step S150: Verify the integrity of the generated stratigraphic model and output the final structured geological model. Preferably, the data standardization operation performed by the seismic data preprocessing module includes normalizing the original seismic amplitude values ​​to the [-1, 1] interval. The signal-to-noise ratio enhancement operation adopts a threshold denoising method based on wavelet transform, with the wavelet basis function selected as Db4, the decomposition layer number as 5, and the soft threshold function defined as... The threshold λ is adaptively calculated based on the noise variance. Furthermore, the initial stratigraphic identification module adopts a multi-attribute fusion stratigraphic extraction strategy. First, it calculates the instantaneous amplitude, instantaneous frequency, and instantaneous phase attributes of the seismic data. Then, it performs dimensionality reduction on the multiple attributes through principal component analysis, retaining the features whose cumulative contribution rate of the first three principal components exceeds 85%. Finally, it uses a region growing algorithm to perform stratigraphic point clustering in the feature space. The region growing criterion is based on the Euclidean distance of the feature vectors, and the distance threshold is set to 1.2 times the standard deviation of the feature space. Furthermore, the fault sensing and adaptive correction module includes a fault detection submodule and a layer correction submodule. The fault detection submodule uses a discontinuity analysis method based on gradient structure tensor to calculate the gradient structure tensor eigenvalues ​​of each sampling point. When the ratio of the minimum eigenvalue to the maximum eigenvalue is less than 0.15, it is determined to be a fault candidate point. Then, morphological operations are used to connect discrete fault points to form a continuous fault line. The layer correction submodule establishes local coordinate systems on both sides of the fault and performs geometric transformations on the layer points using thin-plate spline interpolation to ensure the natural discontinuity and morphological continuity of the layer at the fault. Preferably, the fault sensing and adaptive correction module is connected to generate a continuous three-dimensional stratigraphic surface model based on the corrected stratigraphic point set. This module employs an implicit surface reconstruction method based on radial basis functions, using the corrected stratigraphic point set as a constraint condition to construct a three-dimensional scalar field function. ,in It is a cubic radial basis function. The weighting coefficients wi and polynomial coefficients are determined by solving a system of linear equations constrained by the set of strata points. Finally, the isosurfaces with F(x,y,z)=0 are extracted as three-dimensional strata surfaces using the moving cube algorithm. Furthermore, the quality control and output module performs strata model integrity verification, including checking the topological consistency of the strata surfaces, calculating the average curvature distribution of the strata surfaces to identify abnormal curvature regions, and verifying the degree of matching between the strata and known geological structures. When the area of ​​missing strata regions exceeds 5% of the total area or there are abnormal curvature regions, the strata repair mechanism is automatically triggered, and a surface repair algorithm based on the Poisson equation is used to fill the missing regions. Furthermore, the process of extracting the candidate layer point set in step S120 specifically includes: firstly, calculating the signal envelope along the seismic profile direction and identifying local maxima points as initial candidate layers; then, establishing layer tracking constraints based on seismic wave group characteristics, including phase axis continuity constraints and amplitude consistency constraints; finally, optimizing the layer path through a dynamic programming method, with the objective function considering both signal similarity and path smoothness, and the path smoothness weight coefficient set to 0.3. Preferably, the geometric correction under the influence of the fault line in step S130 is specifically implemented as follows: a buffer zone with a width of 11 sampling points is set on both sides of the identified fault line, an elastic deformation model is applied to the layer points in the buffer zone, the deformation field is obtained by solving the elastic mechanical equilibrium equation, the boundary condition is set to zero displacement at the fault line and continuous displacement at the outer boundary of the buffer zone, and the elastic modulus is dynamically adjusted according to the seismic wave impedance inversion result. Furthermore, the three-dimensional stratum surface generation process in step S140 also includes a surface smoothing optimization operation, which adopts an anisotropic diffusion filtering method. The principal direction of the diffusion tensor is along the normal of the stratum surface, the diffusion coefficient is set to 0.8 in the stratum strike direction and 0.2 in the dip direction, and the number of iterations is controlled within 15 to avoid over-smoothing. In addition, the system also includes a data interface module for exchanging data with external geological databases. It supports input of standard SEGY format seismic data and import of standard LAS format well logging data. The output format supports standard OBJ grid format and custom binary format. It also provides a RESTful API interface for remote calling and integration.

[0007] Compared with the prior art, the present invention has the following beneficial effects: 1. It achieves end-to-end automated processing from raw seismic data to three-dimensional layer models, significantly reducing dependence on external measured data, and is particularly suitable for early exploration stages or areas without wells; 2. Through the embedded fault sensing and adaptive correction mechanism, the problem of stratigraphic misconnection and interruption in complex tectonic areas is effectively solved, and the geometric accuracy and geological rationality of the stratigraphic model are improved. 3. The multi-attribute fusion layer extraction strategy and the physical surface reconstruction method are adopted to ensure the high integrity of layer identification and the continuity of the three-dimensional model, and it has strong adaptability to complex geological structures such as faults and folds. 4. It integrates a comprehensive quality control mechanism, which can automatically detect and repair defects in the layer model, ensuring the reliability and practicality of the output results. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the overall technical architecture of the automatic geological stratum identification and modeling method proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of multi-attribute fusion layer extraction and tomographic sensing adaptive correction in this invention; Figure 3 This is a logical flowchart of the construction and quality control of three-dimensional layered surfaces in this invention. Detailed Implementation

[0009] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the specific embodiments according to the present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments.

[0010] An automatic geological stratum identification and modeling system includes the following components: a seismic data preprocessing module for receiving raw seismic data and performing data standardization and signal-to-noise ratio enhancement operations; a stratum initial identification module connected to the seismic data preprocessing module for extracting a set of candidate stratum points from the preprocessed seismic data; a fault sensing and adaptive correction module connected to the stratum initial identification module for identifying fault structures and performing geometric correction on the candidate stratum points under fault influence; a three-dimensional stratum surface construction module connected to the fault sensing and adaptive correction module for generating a continuous three-dimensional stratum surface model based on the corrected stratum point set; and a quality control and output module connected to the three-dimensional stratum surface construction module for verifying the integrity of the generated stratum model and outputting a final structured geological model. This invention also includes a method for automatic identification and modeling of geological strata, the specific steps of which are as follows: Step S110: Preprocess the raw seismic data, including data standardization and signal-to-noise ratio enhancement operations; Step S120: Extract the candidate set of horizon points from the preprocessed seismic data; Step S130: Identify fault structures and perform geometric correction on candidate stratigraphic points under the influence of faults; Step S140: Generate a continuous three-dimensional stratigraphic surface model based on the corrected stratigraphic point set; Step S150: Verify the integrity of the generated stratigraphic model and output the final structured geological model. Preferably, the data standardization operation performed by the seismic data preprocessing module includes normalizing the original seismic amplitude values ​​to the [-1, 1] interval. The signal-to-noise ratio enhancement operation adopts a threshold denoising method based on wavelet transform, with the wavelet basis function selected as Db4, the decomposition layer number as 5, and the soft threshold function defined as... The threshold λ is adaptively calculated based on the noise variance. The threshold λ is calculated using the formula λ=σ√(2log N), where σ is the noise standard deviation, which is estimated by the amplitude variance of the no-signal area of ​​the seismic data (such as the time window of 100-200ms), and N is the number of sampling points of the seismic trace. Furthermore, the initial stratigraphic identification module employs a multi-attribute fusion stratigraphic extraction strategy. First, it calculates the instantaneous amplitude, instantaneous frequency, and instantaneous phase attributes of the seismic data. Then, it performs dimensionality reduction on the multiple attributes through principal component analysis, retaining features whose cumulative contribution rate of the first three principal components exceeds 85%. Finally, it uses a region growing algorithm to cluster stratigraphic points in the feature space. The region growing criterion is based on the Euclidean distance of the feature vectors, and the distance threshold is set to 1.2 times the standard deviation of the feature space. The distance threshold is also set to 1.2 times the global standard deviation of the feature space. This standard deviation is calculated based on the feature vectors of all candidate stratigraphic points. It has been verified that this threshold can cover 85% of the similarity points of the in-phase axis features. Furthermore, the fault sensing and adaptive correction module includes a fault detection submodule and a layer correction submodule. The fault detection submodule uses a discontinuity analysis method based on gradient structure tensor to calculate the gradient structure tensor eigenvalues ​​of each sampling point. When the ratio of the minimum eigenvalue to the maximum eigenvalue is less than 0.15, it is determined to be a fault candidate point. When the value of the minimum eigenvalue λmin and the maximum eigenvalue λmax, λmin / λmax, is less than the empirical threshold of 0.15 (this threshold is determined through statistical analysis of known fault samples), it is determined to be a fault candidate point. Then, morphological operations are used to connect discrete fault points to form a continuous fault line. The layer correction submodule establishes local coordinate systems on both sides of the fault and performs geometric transformations on the layer points using thin-plate spline interpolation to ensure the natural discontinuity and morphological continuity of the layer at the fault. Preferably, the fault sensing and adaptive correction module is connected to generate a continuous three-dimensional stratigraphic surface model based on the corrected stratigraphic point set. This module employs an implicit surface reconstruction method based on radial basis functions, using the corrected stratigraphic point set as a constraint condition to construct a three-dimensional scalar field function. ,in It is a cubic radial basis function. The weight coefficients wi and polynomial coefficients are determined by solving a system of linear equations constrained by the set of layer points. Finally, the isosurface of F(x,y,z)=0 is extracted as the three-dimensional layer surface by the moving cube algorithm. Furthermore, the quality control and output module performs stratigraphic model integrity verification, including checking the topological consistency of the stratigraphic surface, calculating the average curvature distribution of the stratigraphic surface to identify abnormal curvature areas, verifying the degree of matching between the stratigraphic surface and known geological structures, and automatically triggering the stratigraphic repair mechanism when the area of ​​the missing stratigraphic area exceeds 5% of the total area or there is an abnormal curvature area, using a surface repair algorithm based on the Poisson equation to fill the missing area. Furthermore, the process of extracting the candidate horizon point set in step S120 specifically includes: firstly, calculating the signal envelope along the seismic profile direction and identifying local maxima points as initial candidate horizons; then, establishing horizon tracking constraints based on seismic wave group characteristics, including phase axis continuity constraints and amplitude consistency constraints; finally, optimizing the horizon path using a dynamic programming method, with the objective function defined as follows: Where p is the path point sequence, St is the seismic trace signal at the current location, and S ref The reference signal is used, Corr is the normalized cross-correlation coefficient, α is the similarity weight (set to 1.0), and β is the smoothness weight. The objective function considers both signal similarity and path smoothness, with the path smoothness weighting coefficient β set to 0.3, and preferably, the geometric correction under the influence of the fault line in step S130 is specifically implemented as follows: a buffer zone with a width of 11 sampling points is set on both sides of the identified fault line, an elastic deformation model is applied to the layer points in the buffer zone, the deformation field is obtained by solving the elasticity equilibrium equation, the boundary condition is set to zero displacement at the fault line and continuous displacement at the outer boundary of the buffer zone, and the elastic modulus is dynamically adjusted according to the seismic wave impedance inversion results. Furthermore, the three-dimensional stratum surface generation process in step S140 also includes a surface smoothing optimization operation, which adopts an anisotropic diffusion filtering method. The principal direction of the diffusion tensor is along the normal of the stratum surface, the diffusion coefficient is set to 0.8 in the stratum strike direction and 0.2 in the dip direction, and the number of iterations is controlled within 15 to avoid over-smoothing. In addition, the system also includes a data interface module for exchanging data with external geological databases. It supports input of standard SEGY format seismic data and import of standard LAS format well logging data. The output format supports standard OBJ grid format and custom binary format. It also provides a RESTful API interface for remote calling and integration.

[0011] In the early stages of oil and gas exploration, especially in areas with no or sparse well control, automated geological stratigraphy identification and modeling systems begin to execute complete processing flows. See also... Figure 1 The system first receives raw seismic data acquired in the field through a seismic data preprocessing module. This data typically contains signal distortions caused by environmental noise, instrument acquisition errors, and formation absorption attenuation. Data standardization normalizes the raw seismic amplitude values ​​to the range of -1 to 1, eliminating dimensional differences caused by different acquisition batches or equipment. Signal-to-noise ratio (SNR) enhancement employs a wavelet transform-based threshold denoising method. The wavelet basis function is Db4, and the decomposition layer is five layers. The soft threshold function is defined as the positive part of the sign function multiplied by its absolute value minus the threshold value. The threshold is adaptively calculated based on the noise variance, specifically derived by statistically analyzing regions in the seismic data that do not contain effective signals. The preprocessed seismic data shows an approximately 8 dB increase in SNR, laying the foundation for subsequent layer identification. The SNR improvement is achieved through the formula... Quantification, among which The effective signal variance (calculated using the target layer amplitude). The noise variance (calculated through the no-signal region) is used to compare the data before and after denoising, resulting in an SNR improvement of 8.2 ± 0.5 dB.

[0012] The initial stratigraphic identification module is connected to the seismic data preprocessing module and employs a multi-attribute fusion-based stratigraphic extraction strategy. First, the instantaneous amplitude, instantaneous frequency, and instantaneous phase attributes of the seismic data are calculated. Instantaneous amplitude reflects the reflection intensity at stratigraphic interfaces, instantaneous frequency indicates stratigraphic thickness variations, and instantaneous phase characterizes wavegroup continuity. Then, principal component analysis is used to reduce the dimensionality of the multiple attributes, retaining features where the cumulative contribution rate of the first three principal components exceeds 85%, and eliminating redundant information between attributes. Finally, a region growing algorithm is used to cluster stratigraphic points in the feature space. The region growing criterion is based on the Euclidean distance of the feature vectors, with a distance threshold set to 1.2 times the standard deviation of the feature space. This module outputs a set of candidate stratigraphic points, each point containing its three-dimensional coordinates and corresponding multi-attribute feature vector.

[0013] The fault sensing and adaptive correction module is connected to the initial floor plan identification module and includes a fault detection submodule and a floor plan correction submodule. See also... Figure 2 The fault detection submodule employs a discontinuity analysis method based on gradient structure tensors. It calculates the gradient structure tensor eigenvalues ​​of each sampling point. Points with a ratio of minimum to maximum eigenvalue below 0.15 are identified as candidate fault points. Then, morphological operations are used to connect discrete fault points to form continuous fault lines. Morphological closing operations are also used to connect discrete fault points. A 3×3 rectangular kernel is used as the structural element, and the iteration count is 2. The stratigraphic correction submodule establishes local coordinate systems on both sides of the fault and performs geometric transformations on the stratigraphic points using thin-plate spline interpolation to ensure the natural discontinuity and morphological continuity of the stratigraphic layers at the fault. A buffer zone with a width of eleven sampling points was set on both sides of the identified fault line. An elastic deformation model was applied to the layer points within the buffer zone. The linear elastic deformation model was applied, and the governing equation was the Navierian equation μ∇²u+(λ+μ)∇(∇·u)=0, where u is the displacement field vector, and λ and μ are Lamé constants converted from the elastic modulus. The boundary conditions were set as follows: u=0 at the fault line, and the displacement of the outer boundary of the buffer zone is continuous with the undeformed region. The finite element method was used for numerical solution, with the boundary conditions set as zero displacement at the fault line and continuous displacement of the outer boundary of the buffer zone. The elastic modulus was dynamically adjusted according to the seismic impedance inversion results.

[0014] The 3D stratigraphic surface construction module is connected to the fault sensing and adaptive correction module, employing an implicit surface reconstruction method based on radial basis functions. Using the corrected stratigraphic point set as constraints, a 3D scalar field function is constructed, equal to the weighting coefficients multiplied by a cubic radial basis function plus a linear polynomial. The cubic radial basis function uses distance as the independent variable, and the linear polynomial ensures the affine invariance of the function. The weighting coefficients are determined by solving a system of linear equations. Finally, isosurfaces with zero function values ​​are extracted using the moving cube algorithm as the 3D stratigraphic surface. The surface generation process also includes surface smoothing optimization, employing an anisotropic diffusion filtering method. The principal direction of the diffusion tensor is along the stratigraphic surface normal, the diffusion coefficient is set to 0.8 in the strike direction and 0.2 in the dip direction, and the number of iterations is controlled to within fifteen to avoid over-smoothing.

[0015] The quality control and output module connects to the 3D stratigraphic surface construction module to perform stratigraphic model integrity verification. This includes checking the topological consistency of the stratigraphic surface, calculating the average curvature distribution of the stratigraphic surface to identify abnormal curvature regions, and verifying the degree of matching between the stratigraphic surface and known geological structures. When a stratigraphic missing area is detected to exceed 5% of the total area or an abnormal curvature region exists, the stratigraphic repair mechanism is automatically triggered. A surface repair algorithm based on the Poisson equation is used to fill the missing area: first, Dirichlet boundary conditions are defined at the boundary of the missing area; second, the Poisson equation is established. Where v is the guiding field vector, calculated based on the known surface normal at the boundary; finally, the Poisson equation is solved by the finite element method or finite difference method to obtain the scalar field function f in the missing region, and then the complete surface is reconstructed; the final structured geological model is output, supporting standard OBJ mesh format and custom binary format, and exchanging data with external geological databases through the data interface module, supporting standard SEGY format seismic data input and standard LAS format well logging data import, and providing a RESTful API interface for remote calling and integration.

[0016] In the specific implementation process, step S110 preprocesses the raw seismic data, including data standardization and signal-to-noise ratio (SNR) enhancement. Data standardization unifies seismic data with different acquisition parameters to the same dimension, while SNR enhancement preserves effective signals through wavelet threshold denoising. Step S120 extracts a set of candidate horizon points from the preprocessed seismic data. First, the signal envelope is calculated along the seismic profile direction, and local maxima points are identified as initial candidate horizons. Then, horizon tracking constraints are established based on seismic wave group characteristics, including phase axis continuity constraints and amplitude consistency constraints. Finally, the horizon path is optimized using a dynamic programming method, with the objective function considering both signal similarity and path smoothness, and the path smoothness weight coefficient set to 0.3.

[0017] Step S130 identifies fault structures and performs geometric correction on candidate stratigraphic points under fault influence. Fault detection is achieved by analyzing the discontinuities in seismic data, and geometric correction establishes elastic deformation fields on both sides of the fault to ensure reasonable stratigraphic discontinuity at the fault. Step S140 generates a continuous three-dimensional stratigraphic surface model based on the corrected stratigraphic point set. An implicit surface reconstruction method is used to construct the continuous surface, and then anisotropic diffusion filtering is used to optimize the surface quality. Step S150 verifies the integrity of the generated stratigraphic model and outputs the final structured geological model. The verification includes surface topology checking, curvature analysis, and geological rationality assessment. The output format meets industry standard requirements.

[0018] In the application of fine-grained characterization of oil and gas reservoirs in complex structural areas, the automatic geological horizon identification and modeling system optimizes processing for complex geological conditions such as steep structures and thrust faults. The seismic data preprocessing module addresses the seismic wave scattering phenomenon unique to complex structural areas by adding directional filtering to standard wavelet denoising, focusing on preserving signal characteristics along the dip direction of strata. The initial horizon identification module introduces curvature and coherence attributes during multi-attribute fusion to enhance the identification of fold and fault boundaries. The distance threshold of the region growing algorithm is dynamically adjusted according to the local structural complexity, using 1.2 standard deviations in structurally simple areas and tightening to 0.8 standard deviations in structurally complex areas.

[0019] The fault sensing and adaptive correction module, designed for the unique hanging wall and footwall stacking relationships of thrust faults, adds a fault plane attitude analysis function to the fault detection submodule. This function calculates geometric parameters such as fault strike and dip by fitting the spatial distribution of points on the fault plane. The horizon correction submodule employs a dual coordinate system transformation strategy. First, a local coordinate system is established on the footwall to correct horizon points. Then, the correction results are transformed to the global coordinate system through coordinate rotation and translation, ensuring the correct spatial correspondence of horizons on the hanging wall and footwall of the thrust fault. The buffer width is adaptively adjusted according to the magnitude of the fault displacement. Faults with displacement greater than 50 milliseconds use a 15-sampling-point buffer, while faults with smaller displacements use a 7-sampling-point buffer.

[0020] The 3D stratigraphic surface construction module introduces tectonic trend constraints during radial basis function reconstruction, incorporating regional geological tectonic background knowledge as a soft constraint into the surface reconstruction process to prevent the surface from overfitting noise points and deviating from geological regularities. The quality control and output module adds a stratigraphic contact relationship verification function to check whether the contact relationships such as overlap and truncation between different stratigraphic layers conform to geological regularities. When abnormal contact is detected, a manual interactive verification process is automatically initiated. The data interface module is expanded to support commonly used data formats in geological modeling software, achieving seamless integration with mainstream geological modeling platforms.

[0021] At the methodological execution level, step S120, the extraction of candidate stratigraphic points, optimizes the objective function of the dynamic programming algorithm for complex tectonic regions. It adds a structural consistency term to the signal similarity and path smoothness criteria to ensure that the tracked stratigraphic paths conform to regional tectonic trends. Step S130, geometric correction under the influence of faults, introduces a multi-scale analysis strategy. First, it determines the overall control effect of faults on stratigraphic levels at a large scale, and then refines the local morphology of stratigraphic levels at a small scale. Step S140, the generation of 3D stratigraphic surfaces, employs a multi-resolution reconstruction strategy. It first constructs the overall morphology of the stratigraphic level based on sparse point sets, and then gradually incorporates dense point sets to refine local features, balancing computational efficiency and modeling accuracy. Step S150, integrity verification adds a stratigraphic closure error check to ensure that the geometric closure of the stratigraphic model in 3D space meets geological modeling requirements.

Claims

1. An automatic geological stratigraphic identification and modeling system, characterized in that: The system comprises the following components: seismic data preprocessing module, initial horizon identification module, fault sensing and adaptive correction module, three-dimensional horizon surface construction module, and quality control and output module; The seismic data preprocessing module receives raw seismic data and performs data standardization and signal-to-noise ratio (SNR) enhancement operations. Data standardization includes normalizing the raw seismic amplitude values ​​to the [-1, 1] interval. SNR enhancement employs a wavelet transform-based threshold denoising method with Db4 as the wavelet basis function and 5 decomposition layers. The soft threshold function is defined as... The threshold λ is adaptively calculated based on the noise variance. The initial stratigraphic identification module is connected to the seismic data preprocessing module and is used to extract a set of candidate stratigraphic points from the preprocessed seismic data. This module adopts a multi-attribute fusion stratigraphic extraction strategy. First, it calculates the instantaneous amplitude, instantaneous frequency, and instantaneous phase attributes of the seismic data. Then, it performs dimensionality reduction processing on the multiple attributes through principal component analysis, retaining the features whose cumulative contribution rate of the first three principal components exceeds 85%. Finally, it uses a region growing algorithm to perform stratigraphic point clustering in the feature space. The region growing criterion is based on the Euclidean distance of the feature vectors, and the distance threshold is set to 1.2 times the standard deviation of the feature space. The fault sensing and adaptive correction module is connected to the initial stratigraphic identification module. It is used to identify fault structures and perform geometric correction on candidate stratigraphic points under the influence of faults. This module includes a fault detection submodule and a stratigraphic correction submodule. The fault detection submodule uses a discontinuity analysis method based on gradient structure tensor to calculate the gradient structure tensor eigenvalue of each sampling point. When the ratio of the minimum eigenvalue to the maximum eigenvalue is less than 0.15, it is determined to be a candidate fault point. Then, discrete fault points are connected to form a continuous fault line through morphological operations. The stratigraphic correction submodule establishes local coordinate systems on both sides of the fault, performs geometric transformation on the stratigraphic points through thin plate spline interpolation, and sets a buffer with a width of 11 sampling points on both sides of the identified fault line. An elastic deformation model is applied to the stratigraphic points in the buffer. The three-dimensional stratigraphic surface construction module is connected to the fault sensing and adaptive correction module. It generates a continuous three-dimensional stratigraphic surface model based on the corrected stratigraphic point set. This module employs an implicit surface reconstruction method based on radial basis functions, using the corrected stratigraphic point set as a constraint condition to construct a three-dimensional scalar field function. ,in It is a cubic radial basis function. The weight coefficients wi and polynomial coefficients are determined by solving a system of linear equations constrained by the set of layer points. Finally, the isosurface of F(x,y,z)=0 is extracted as the three-dimensional layer surface by the moving cube algorithm. The quality control and output module is connected to the three-dimensional stratigraphic surface construction module. It is used to verify the integrity of the generated stratigraphic model and output the final structured geological model. The integrity verification includes checking the topological consistency of the stratigraphic surface, calculating the average curvature distribution of the stratigraphic surface to identify abnormal curvature areas, and verifying the degree of matching between the stratigraphic surface and known geological structures. When the area of ​​the missing stratigraphic area exceeds 5% of the total area or there is an abnormal curvature area, the stratigraphic repair mechanism is automatically triggered.

2. The automatic geological stratum identification and modeling system according to claim 1, characterized in that: The initial stratigraphic identification module includes the following steps for extracting candidate stratigraphic points: First, the signal envelope is calculated along the seismic profile direction, and local maxima are identified as initial stratigraphic candidates. Then, stratigraphic tracking constraints are established based on seismic wave group characteristics, including phase axis continuity constraints and amplitude consistency constraints. Finally, the stratigraphic path is optimized using a dynamic programming method, with the objective function considering both signal similarity and path smoothness, and the path smoothness weight coefficient is set to 0.

3.

3. The automatic geological stratigraphic identification and modeling system according to claim 1, characterized in that: In the fault sensing and adaptive correction module, the geometric correction under the influence of faults is specifically implemented as follows: a buffer zone with a width of 11 sampling points is set on both sides of the identified fault line. An elastic deformation model is applied to the layer points in the buffer zone. The deformation field is obtained by solving the elastic mechanical equilibrium equation. The boundary conditions are set so that the displacement at the fault line is zero while the displacement at the outer boundary of the buffer zone is continuous. The elastic modulus is dynamically adjusted according to the seismic wave impedance inversion results.

4. The automatic geological stratum identification and modeling system according to claim 1, characterized in that: In the three-dimensional stratum surface construction module, the three-dimensional stratum surface generation process also includes surface smoothing optimization operation, which adopts an anisotropic diffusion filtering method. The main direction of the diffusion tensor is along the normal of the stratum surface, the diffusion coefficient is set to 0.8 in the stratum strike direction and 0.2 in the dip direction, and the number of iterations is controlled within 15.

5. The automatic geological stratigraphic identification and modeling system according to claim 1, characterized in that: In the quality control and output module, the layer repair mechanism uses a surface repair algorithm based on the Poisson equation to fill the missing regions. This algorithm establishes a continuous surface extension at the boundary of the missing region by solving the Poisson equation, thus maintaining the geometric continuity and topological consistency of the layer surface.

6. The automatic geological stratigraphic identification and modeling system according to any one of claims 1-5, characterized in that: The system also includes a data interface module for exchanging data with external geological databases. It supports input of standard SEGY format seismic data and import of standard LAS format well logging data. The output format supports standard OBJ grid format and custom binary format. It also provides a RESTful API interface for remote calling and integration.

7. A method for automatic identification and modeling of geological strata, characterized in that: The method includes the following steps: S110 preprocesses the raw seismic data, including data standardization and signal-to-noise ratio enhancement. Data standardization normalizes the raw seismic amplitude values ​​to the [-1,1] interval, and signal-to-noise ratio enhancement uses a threshold denoising method based on wavelet transform. S120 extracts candidate point sets of horizons from preprocessed seismic data, adopts a multi-attribute fusion horizon extraction strategy, calculates the instantaneous amplitude, instantaneous frequency and instantaneous phase attributes of seismic data, performs dimensionality reduction processing on multiple attributes through principal component analysis, retains features with a cumulative contribution rate of more than 85% for the first three principal components, and uses a region growing algorithm to perform horizon point clustering in the feature space. S130, identify fault structures and perform geometric correction on candidate strata under the influence of faults. The fault structure is identified by the discontinuity analysis method based on gradient structure tensor. The strata are geometrically transformed by thin plate spline interpolation method. Buffer zones are set on both sides of the fault line and elastic deformation model is applied. S140: A continuous three-dimensional stratigraphic surface model is generated based on the corrected stratigraphic point set. A three-dimensional scalar field function is constructed using an implicit surface reconstruction method based on radial basis functions. The isosurface is extracted as the three-dimensional stratigraphic surface through the moving cube algorithm. S150 verifies the integrity of the generated stratigraphic model and outputs the final structured geological model. The verification includes checking the topological consistency of the stratigraphic surfaces, calculating the average curvature distribution of the stratigraphic surfaces, and verifying the degree of matching between the stratigraphic surfaces and known geological structures.

8. The method for automatic identification and modeling of geological strata according to claim 7, characterized in that, Step S140 also includes a surface smoothing optimization operation, in which an anisotropic diffusion filtering method is used to smooth the generated three-dimensional layer surface. The principal direction of the diffusion tensor is along the normal of the layer surface, the diffusion coefficient is set to 0.8 in the strike direction and 0.2 in the dip direction, and the number of iterations is controlled to be within 15.

Images