A method for delineating boundaries of a glutenite fan body based on substrate fracture

Abstract

The application provides a sand and gravel fan body boundary delineation method based on a basement fracture, which comprises the following steps: S1, collecting remote sensing image data and geological survey data of the sand and gravel fan body, and obtaining measurement data of the sand and gravel fan body; S2, establishing a full-area sand and gravel body isochronous stage model according to the remote sensing image data and the geological survey data, and determining preliminary characteristics; S3, obtaining boundary morphology and position information of the sand and gravel fan body based on basement fracture characteristics; S4, optimizing and delineating the boundary morphology and position information of the sand and gravel fan body, and obtaining quantitative boundary characteristics of the fan body; S5, dividing the sand and gravel fan body into multiple clustering clusters according to the quantitative boundary characteristics, and determining the boundary of each clustering cluster; S6, predicting the oil-bearing height of each sand and gravel fan body; and S7, presenting the quantitative boundary characteristics to a user. More reliable geological models and reservoir prediction results are provided.

Description

Technical Field

[0001] This invention relates to the field of petroleum exploration and development, and in particular to a method for delineating the boundaries of sandstone and conglomerate fan bodies based on basement fractures. Background Technology

[0002] With the increasingly severe global oil resource situation, higher demands are placed on improving exploration success rates, reducing exploration risks, and increasing exploration efficiency under the current circumstances of efficient exploration and development. For geological structures with complex structures and large strata dip angles, such as the sandstone and conglomerate fan bodies in the steep slope zone of the northern Dongying Depression, the accurate depiction of the distribution of these fan bodies, controlled by the Chennan basement fault, directly affects the oil-bearing zones of the sandstone and conglomerate reservoirs, thus impacting exploration and development efficiency. The previously understood theory of "trench-fan correspondence" needs further refinement and in-depth study in the increasingly challenging exploration conditions of the old eastern regions.

[0003] In addition to the large gullies, there are also smaller gullies in the horizontal plane. Similarly, sandstone and conglomerate fans are superimposed and developed between large gullies, between large gullies and smaller gullies, and between smaller gullies. Vertically, due to the episodic evolution of the basement faults, the multi-stage sandstone and conglomerate fans also exhibit segmented and differentiated development characteristics. Through investigation, it was found that current research on the Chennan basement fault mainly focuses on the overall analysis of the general patterns of the Chennan fault's horizontal and vertical characteristics, and has not yet developed a quantitative and specific identification method for delineating the boundaries of sandstone and conglomerate fans based on the basement faults.

[0004] Currently, the traditional, broad-based study of basement fracture characteristics can no longer meet the needs of highly mature exploration areas, such as the old eastern regions. Summary of the Invention

[0005] In view of the above problems, the present invention is proposed to provide a method for delineating the boundaries of sandstone and conglomerate fans based on basement fractures, which overcomes or at least partially solves the above problems.

[0006] According to one aspect of the present invention, a method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture is provided, the boundary delineation method comprising:

[0007] Step S1: Collect remote sensing image data and geological survey data of the sandstone and conglomerate fan, and obtain measurement data of the sandstone and conglomerate fan;

[0008] Step S2: Based on the remote sensing image data and geological survey data, establish a phase model of the sandstone and conglomerate bodies in the whole area and determine the preliminary characteristics;

[0009] Step S3: Based on the basement fracture characteristics, obtain the boundary morphology and location information of the conglomerate fan.

[0010] Step S4: Optimize and characterize the boundary morphology and location information of the sandstone and conglomerate fan to obtain the quantitative boundary characteristics of the fan;

[0011] Step S5: Divide the sandstone and conglomerate fan into multiple clusters based on quantitative boundary characteristics, and determine the boundary of each cluster;

[0012] Step S6: Establish a pressure sensitivity model for sandstone and conglomerate reservoirs to predict the oil-bearing height of each phase of sandstone and conglomerate fan bodies;

[0013] Step S7: Present the quantitative boundary features to the user.

[0014] Optionally, step S2: establishing a phase model of the entire area's sandstone and conglomerate bodies based on the remote sensing image data and geological survey data, and determining preliminary characteristics specifically includes:

[0015] The well-seismic fine calibration was adopted, combined with actual drilled well core samples, logging and analysis data, and based on the aforementioned remote sensing image data and geological survey data;

[0016] By using a comparative analysis of well-connected skeleton profiles, a phase model of the sandstone and conglomerate bodies in the entire area was established, and preliminary characteristics were determined.

[0017] Optionally, step S2: establishing a phase model of the entire area's sandstone and conglomerate bodies based on the remote sensing image data and geological survey data, and determining preliminary characteristics specifically includes:

[0018] Determine the variable n to be analyzed i , i = 1, 2, ... a, including groundwater content and the abundance of petroleum mineral resources;

[0019] Determine multiple levels of multiple factors r t-p t = 1, 2, ..., b, where multiple factors include topography, geological structure, and lithology, and p ∈ {the set of level values ​​corresponding to the factors}.

[0020] The total variance is SS tatal =SS between +SS within SS total This represents the variance of all data.

[0021] Within-group variance Indicates that r is being analyzed t-p The variance of the set of data, where For the rth t-p Observations of the set of data For the first The average of the group data;

[0022] Between-group variance This represents the variance among multiple groups. For the rtht-p Group sample size The average of all data;

[0023] F-value = MSB / MSW, which represents the ratio of the mean square MSB between groups to the mean square MSW within groups. The larger the F-value, the more significant the difference.

[0024] Optionally, step S3: obtaining the boundary morphology and location information of the conglomerate fan based on the basement fracture characteristics specifically includes:

[0025] Based on the phase model of the sandstone and conglomerate bodies in the whole region, profiles were extracted for each phase.

[0026] Fault lines were extracted based on velocity differences;

[0027] An information transformation neural network, including an input layer, a transformation layer, an optimization layer, and an output layer, is constructed to convert collected data information into image information. At the same time, a semi-supervised learning model based on transform is constructed to identify and analyze the boundaries of conglomerate fan bodies.

[0028] Optionally, step S3: obtaining the boundary morphology and location information of the conglomerate fan based on the basement fracture characteristics further includes:

[0029] The input layer and transformation layer are fully connected. The transformation layer transforms the data information. The current data information has N features, specifically:

[0030]

[0031] Where sr1 represents the input of the transformation layer, ω 21 Q represents the association weight between the input layer and the transformation layer. j Let θ represent the j-th data feature input to the input layer, θ represent the bias of the transformation layer, f represent the activation function of the transformation layer, T represent the bias, K represent the transformation factor, and sc1 represent the output of the transformation layer. f represents the learning factor. z Let z represent the z-th transformation object, and γ represent the stability factor;

[0032] The transformation layer passes the processing results to the optimization layer for optimization. The optimization layer's calculation process is: sr2 = ω 32 sc1+θ'

[0033]

[0034] Where sr2 represents the input to the optimization layer, ω 32 The weights represent the correlation between the transformation layer and the optimization layer, θ' represents the bias of the optimization layer, β represents the control parameter, and ρ represents the preset adjustment parameter; j ≠ j'; f j'This represents the transformation object corresponding to the j'th data information feature.

[0035] Optionally, step S4: optimizing and characterizing the boundary morphology and location information of the conglomerate fan to obtain the quantitative boundary features of the fan specifically includes:

[0036] The sandstone and conglomerate fan was divided into multiple units of uniform size and shape using gridding or voxelization methods;

[0037] Calculate the distance and angle between each element and the base fracture, and use the distance and angle between each element and the base fracture as the element's characteristic value.

[0038] Optionally, step S5: dividing the sandstone and conglomerate fan into multiple clusters based on quantitative boundary characteristics, and determining the boundary of each cluster specifically includes:

[0039] For the boundary of each cluster, a fitting algorithm is used to fit a curve to obtain a model of the cluster boundary;

[0040] The cluster boundaries were optimized and corrected to obtain the final sandstone and conglomerate fan body boundaries.

[0041] Optionally, after obtaining the final sandstone and conglomerate fan boundary, the method further includes: performing an error analysis on the boundary.

[0042] Optionally, the error analysis of the boundary specifically includes: boundary morphology, boundary change trend, and boundary location.

[0043] Optionally, step S6: establishing a pressure sensitivity model for sandstone and conglomerate reservoirs and predicting the oil-bearing height of each phase of the sandstone and conglomerate fan specifically includes:

[0044] A pressure sensitivity model for sandstone and conglomerate reservoirs was established. Based on information and geological structure, two suitable models were selected and combined to obtain comprehensive prediction results.

[0045] The exponential model is as follows: Where k is the penetration rate, and k_0 is the base penetration rate. It represents porosity, and m is a parameter of the exponential model;

[0046] The elastic modulus E and Poisson's ratio v of sandstone and conglomerate reservoirs were calculated based on the Biot-Gassmann model.

[0047] Poisson's ratio sensitivity index O for sandstone and conglomerate reservoirs; permeability sensitivity index I for sandstone and conglomerate reservoirs;

[0048] A regression model was established using Poisson's ratio sensitivity index and permeability sensitivity index as independent variables, and reservoir productivity or reserves index as dependent variables.

[0049] Optionally, step S7: presenting the quantitative boundary features to the user specifically includes:

[0050] Using 3D visualization technology, quantitative boundary features are presented to users, enabling the analysis and comparison of the internal structure and properties of the sector.

[0051] This invention provides a method for delineating the boundaries of conglomerate fans based on basement faults. The method includes: Step S1: acquiring remote sensing image data and geological survey data of the conglomerate fan, and obtaining measurement data of the conglomerate fan; Step S2: establishing a phased model of the entire conglomerate body based on the remote sensing image data and geological survey data, and determining preliminary characteristics; Step S3: obtaining the boundary morphology and location information of the conglomerate fan based on basement fault characteristics; Step S4: optimizing and delineating the boundary morphology and location information of the conglomerate fan to obtain quantitative boundary characteristics of the fan; Step S5: dividing the conglomerate fan into multiple clusters based on the quantitative boundary characteristics, and determining the boundary of each cluster; Step S6: establishing a pressure sensitivity model for conglomerate reservoirs to predict the oil-bearing height of each phase of the conglomerate fan; Step S7: presenting the quantitative boundary characteristics to the user. This method accurately delineates the morphology and structural features of conglomerate fans, providing more reliable geological models and reservoir prediction results.

[0052] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 A flowchart illustrating a method for delineating the boundary of a sandstone and conglomerate fan based on basal fracture, provided for an embodiment of the present invention;

[0055] Figure 2 This is a diagram showing the distribution and migration characteristics of sandstone in different phases of the conglomerate rock formations of this invention.

[0056] Figure 3 This is a thickness map of the Sha-3 lower reservoir in a certain region according to a specific embodiment of the present invention;

[0057] Figure 4 This is a sedimentary facies diagram of the Sandy Land III in a certain region, according to a specific embodiment of the present invention.

[0058] Figure 5 This is a north-south seismic profile of a well in a specific embodiment of the present invention. Detailed Implementation

[0059] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0060] The terms "comprising" and "having," and any variations thereof, in the specification, embodiments, claims, and drawings of this invention are intended to cover non-exclusive inclusion, such as including a series of steps or units.

[0061] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0062] like Figure 1 and Figure 2 As shown, this embodiment provides a method for quantitative characterization of sandstone and conglomerate fan body boundaries based on basement fracture control, including the following steps:

[0063] S1. Acquire high-resolution remote sensing imagery and geological survey data of the sandstone and conglomerate fan, including data on fault distribution, lithology, geological structure, and physical properties of the sandstone and conglomerate fan; acquire measurement data of the sandstone and conglomerate fan, including topographic data, geological and geophysical data, and borehole data; and input the data into a computer model. In specific implementation applications, when acquiring data, obtain a thickness map of the lower Shale Sediment 3 reservoir in a certain area (e.g., Figure 3 (as shown) and the Lower Sedimentary Facies Diagram of Sha-3 (as shown) Figure 4 As shown), there is also a north-south seismic profile of a well (as shown). Figure 5 As shown in the figure, this provides strong support and foundation for subsequent data analysis and modeling processes.

[0064] S2. Through fine calibration by well seismic testing, combined with analysis and testing data such as core samples from actual drilled wells and logging data, and based on the data acquisition results, a phase model of the sandstone and conglomerate bodies in the whole area is established by comparative analysis of the well skeleton profiles, and features are automatically extracted.

[0065] A phase-series model of the sandstone and conglomerate bodies in the entire region was established, as detailed below:

[0066] a. Data collected through geological surveys, remote sensing images, and surveying and exploration, including stratigraphic distribution, lithology, thickness, slope, dip angle, faults, folds, toughness, and slip surfaces, are organized, optimized, screened, and standardized to eliminate outliers, noise, and errors in the data.

[0067] b. Based on geological data and geological principles, establish a stratigraphic interface model G, including determining the thickness, dip angle, and strike of strata; based on collected rock samples and relevant seismic data, establish a lithological model H, including the distribution, type, and thickness of lithology; based on seismic data and fault analysis, establish a structural model J, including folds, faults, and strata dip angles; based on geophysical models and geological data, establish an attribute model K, including groundwater level, temperature, and pressure; then, integrate the stratigraphic interface model G, lithological model H, structural model J, and attribute model K to construct a three-dimensional geological model.

[0068] c. By determining the variable n to be analyzed i i = 1, 2, ..., a, for example: groundwater content, abundance of petroleum mineral resources, etc.; determine different levels r of different factors. t-p Let t = 1, 2, ..., b, where different factors include topography, geological structure, lithology, etc., and p ∈ {the set of level values ​​corresponding to the factors}. Using analysis of variance, assume the total variance SS tatal =SS between +SS within SS tatal This represents the variance of all data.

[0069] Within-group variance Indicates that r is being analyzed t-p The variance of the set of data, where For the rth t-p Observations of the set of data For the rth t-p Mean of group data; between-group variance This represents the variance between different groups. For the rth t-p Group sample size The average of all data;

[0070] F-value = MSB / MSW, which represents the ratio of the mean square MSB between groups to the mean square MSW within groups. The larger the F-value, the more significant the difference.

[0071] The spatial distribution characteristics of the phase models of sandstone and conglomerate bodies are determined through a series of steps.

[0072] d. Then, use existing technologies to evaluate and optimize the phase models of sandstone and conglomerate bodies in the whole region, determine the accuracy and reliability of the models, and improve the accuracy and predictive ability of the models.

[0073] This step establishes phase models of sandstone and conglomerate bodies across the region, enabling a more refined division and interpretation of geological history and a better understanding of geological evolution. It helps identify geological structural features at different stages, such as faults and folds, facilitating a comprehensive analysis of regional geological structures. Furthermore, it provides a better understanding of the formation and distribution patterns of sandstone and conglomerate bodies within the region, thereby improving the accuracy of sandstone and conglomerate body predictions. Moreover, it offers guidance on the migration and variation patterns of groundwater, rocks, and other materials within the region, providing a scientific basis for engineering exploration and data and methodological support for in-depth research in related fields. This is of great significance for regional economic and social development and resource utilization.

[0074] S3. Based on the characteristics of the basement fracture and combined with three-dimensional seismic imaging technology, the boundary identification and analysis of the inside and outside of the fan body are realized, and the boundary morphology and location information of the sandstone and conglomerate fan body are obtained.

[0075] Based on the acquired seismic data, the reflecting surface of the basement was determined. Then, according to the phase model of the conglomerate mass in the entire region, profiles were extracted for each phase to obtain profile data. The obtained profile data was then used for fault extraction. Fault extraction yielded boundary information of the internal and external parts of the fan-shaped rock mass, which helps to establish a more accurate geological model, as detailed below:

[0076]

[0077] in, The values ​​represent the velocity difference: t1 is the time it takes for the seismic wave to travel from the source to the underground rock, t2 is the time it takes for the seismic wave to travel from the source to the ground after being reflected by the underground rock, and t2-t1 represents the time it takes for the seismic wave to travel inside the rock.

[0078] By extracting fault lines based on velocity differences, interpolation methods can be used to convert these velocity differences into fault line location information, thereby further determining the fault's strike, dip angle, and other related characteristics. Fault extraction provides crucial constraints during model construction and parameter adjustment, improving the model's accuracy and reliability. It also helps identify potential geological hazards and provides important references for regional development and engineering construction, aiding in further exploration well drilling.

[0079] A semi-supervised learning model based on transform is constructed to achieve boundary recognition and analysis. First, the collected data information needs to be converted into image information. An information transformation neural network is constructed, including an input layer, a transformation layer, an optimization layer, and an output layer. Assuming the current data information has N features, the input layer and the transformation layer are fully connected. The transformation layer transforms the data information using the following formula:

[0080]

[0081] Where sr1 represents the input of the transformation layer, ω 21 Q represents the association weight between the input layer and the transformation layer. j Let θ represent the j-th data feature input to the input layer, θ represent the bias of the transformation layer, f represent the activation function of the transformation layer, T represent the bias, K represent the transformation factor, and sc1 represent the output of the transformation layer. f represents the learning factor. z Let z represent the z-th transformation object, and γ represent the stability factor.

[0082] The transformation layer passes the processing results to the optimization layer; that is, the output of the transformation layer is the input of the optimization layer. The optimization layer optimizes the transformation results to ensure the accuracy and usability of the obtained image data. The calculation process of the optimization layer is as follows:

[0083] sr2=ω 32 sc1+θ

[0084]

[0085] Where sr2 represents the input to the optimization layer, ω 32 The weights represent the correlation between the transformation layer and the optimization layer, θ' represents the bias of the optimization layer, β represents the control parameter, and ρ represents the preset adjustment parameter; j ≠ j'; f j' This represents the transformation object corresponding to the j'-th data information feature. The optimization layer then passes the final optimization information to the output layer, which will output the required image information, providing a foundation for the transform-based semi-supervised learning model.

[0086] The obtained geological image data is labeled and converted into a format suitable for model processing. Image enhancement techniques are used to improve image quality and readability, including rotation, translation, scaling, and cropping. Boundary annotation tools are also used to label the boundaries of sandstone and conglomerate fans for semi-supervised learning.

[0087] The VIT model is used to extract image features, and multiple Transformer modules are used to extract the feature representation of the image, where each Transformer module contains a multi-head self-attention mechanism and a fully connected network.

[0088] First, define the input image as x∈R. H×W×C Where H, W, and C represent the height, width, and number of channels of the image, respectively;

[0089] The first step is to extract representations of image features, as follows:

[0090] y tr(x)=cov1(Norm1(ReLU(w1*x+b1)))

[0091] y tr (x)=cov2(Norm2(ReLU(w2*y tr (x)+b2)))

[0092] y tr (x)=cov3(Norm3(ReLU(w3*y tr (x)+b3)))

[0093] Among them, y tr (x) represents the image feature representation; cov1 represents convolutional layer 1, cov2 represents convolutional layer 2, and cov3 represents convolutional layer 3; Norm1 represents the batch normalization layer, which normalizes the input data from layer 1; Norm2 represents the normalization layer for the input data from layer 2; Norm3 represents the normalization layer for the input data from layer 3; ReLU represents the nonlinear activation function, used to increase the nonlinearity of the network; w1 and b1 represent the weights and biases of convolutional layer 1, respectively; w2 and b2 represent the weights and biases of convolutional layer 2, respectively; and w3 and b3 represent the weights and biases of convolutional layer 3, respectively.

[0094] The second step is to perform feature fusion, as shown below:

[0095]

[0096] Among them, y ne (x1,x2,…,x n ) represents the feature representation after feature fusion; w n b represents the weight values ​​of the feature fusion layer; n This represents the bias term of the feature fusion layer.

[0097] The third step is to output the classification results, as shown below:

[0098] y ta (x) = softmax(w) out *y ne (y tr (x))+b out )

[0099] Among them, y ta (x) represents the result of classifying the image; softmax is a non-linear function that transforms the model's output into a probability distribution; w out b represents the weight values ​​of the output layer. out This represents the bias term of the output layer.

[0100] Then, a semi-supervised learning method is used to train the model. In semi-supervised learning, a self-training method is used to train the VIT model. Semi-supervised learning includes two types of data: labeled data and unlabeled data. Labeled data is used to calculate the loss function, and unlabeled data is used to improve the model's generalization ability.

[0101] The specific formula is as follows:

[0102]

[0103] L tat =L sup +αL unsup

[0104] Among them, L sup p represents the cross-entropy loss of labeled data. ij L represents the probability that the true label of the i-th sample is j; unsup L represents the unlabeled maximum prediction loss function; tat The overall loss function is represented by α; α is the hyperparameter used to balance the proportion of labeled and unlabeled data; N represents the total number of samples in the labeled data; M represents the number of categories in the unlabeled data. unlab This represents the total number of samples in the unlabeled data.

[0105] When using a trained VIT model to identify and analyze the boundaries of sandstone and conglomerate fans, methods such as sliding window and probabilistic image semantic segmentation are employed to achieve accurate boundary identification and analysis.

[0106] This invention first constructs an information transformation neural network to convert data information into image information, directly presenting the characteristics and distribution of the data. Based on the image data, a more effective model is built, thereby improving model performance and accuracy. Furthermore, a transform-based semi-supervised learning model is used to identify and analyze the boundaries inside and outside sectors, utilizing unlabeled data to enhance the model's generalization ability, thus improving classification accuracy. Simultaneously, the transform-based method further enhances data diversity and robustness, thereby further improving model performance. The model is trained using a large amount of unlabeled data, reducing labeling costs. This method fully utilizes all data, thereby increasing data utilization and improving model performance.

[0107] S4. The boundary morphology and location information of the sandstone and conglomerate fan are optimized and characterized to obtain the quantitative boundary features of the fan:

[0108] The conglomerate fan is divided into multiple uniformly sized and shaped elements, implemented using methods such as mesh generation or voxelization. For each element, its distance and angle from the basement fault are calculated and used as the element's characteristic values: a. Connect the center point of the element to the center point of the basement fault and calculate the length, which is the distance between the elements; b. Calculate the horizontal and vertical angles between the center point of the element and the basement fault, which are used as the element's angular characteristics.

[0109] For each element, its distance and angle from the base fracture are used as its feature values, resulting in a dataset containing the feature values ​​of all elements. The feature values ​​of each element are then normalized to values ​​between 0 and 1 to facilitate subsequent cluster analysis.

[0110] S5. Using the eigenvalues ​​of the units, a clustering algorithm is used to divide the sandstone and conglomerate fan into multiple clusters, and the boundaries of each cluster are determined.

[0111] Algorithms such as spectral clustering, K-means clustering, hierarchical clustering, and density clustering are used to divide the units into multiple clusters. The number and shape of the clusters need to be selected according to the actual situation.

[0112] In the spectral clustering algorithm, assuming there are n units and each unit has m eigenvalues, the unit feature matrix X is an n×m matrix, where the i-th row represents the m eigenvalues ​​of the i-th unit, i.e.:

[0113] X = [x1, x2, ..., x m ;x2,x2,...,x 2m ;...;x n1 ,x n2 ,...,x nm ]

[0114] The unit similarity matrix W is obtained by calculating the Euclidean distance or cosine similarity between units, where W(i,j) represents the similarity between the i-th unit and the j-th unit.

[0115] The formula for a symmetric normalized Laplace matrix is: L=D^(-1 / 2)*W*D^(-1 / 2), where D is the degree matrix and D(i,i) represents the degree of the i-th element.

[0116] The eigenvalue decomposition of the symmetric normalized Laplace matrix L is: L*F=F*Λ, where F is the eigenvector matrix, Λ is a diagonal matrix, and the elements on the diagonal are the eigenvalues.

[0117] The k eigenvectors with the smallest eigenvalues ​​are selected to form a new matrix F', i.e., F' = [f1, f2, ..., fk]. The row vectors in the new matrix F' are then subjected to k-means analysis to obtain the clusters. Finally, the clustering results are used to determine the boundaries of each cluster.

[0118] For the boundary of each cluster, a fitting algorithm is used to fit its curve to obtain the model of the cluster boundary; then the cluster boundary is optimized and corrected to obtain the final sandstone and conglomerate fan boundary.

[0119] For the boundaries of each cluster, algorithms such as multinomial fitting and spline interpolation are used to fit them to facilitate subsequent data analysis.

[0120] When fitting cluster boundaries using fitting algorithms, some data points that deviate significantly from the actual boundaries may appear and need to be removed. There may be small local fluctuations on the cluster boundaries. In order to make the cluster boundaries smoother, average filtering or Gaussian filtering is used to smooth the boundary curves. Since there may be some overlapping areas between different clusters, it is necessary to connect the boundaries of adjacent clusters to obtain the complete sandstone and conglomerate fan body boundaries.

[0121] Error analysis of the boundary is also required, including aspects such as boundary morphology, boundary change trend, and boundary location.

[0122] S6. Based on the oil and gas migration distance and mode, and combined with the transport capacity of sandstone bodies in the conglomerate framework, a pressure sensitivity model for conglomerate reservoirs was established to obtain the pressure coefficient corresponding to each conglomerate fan. Taking into account the influence of basement fault slope, source supply size, reservoir thickness, sedimentary facies type, effective reservoir sensitivity attributes, physical property changes, oil and gas migration distance, sandstone transport capacity, and pressure sensitivity parameters, the quantitative prediction of the oil-bearing height of each conglomerate fan was finally achieved.

[0123] This can be understood as determining the oil and gas migration distance in conglomerate reservoirs, including the migration distance in the horizontal and vertical directions, as well as the contributions of fractures and pores; and studying the transport capacity of the sandstone skeleton sand bodies through parameters such as porosity, permeability, and pore connectivity, which are obtained through laboratory testing or seismic exploration.

[0124] By establishing a pressure sensitivity model for sandstone and conglomerate reservoirs and combining two suitable models based on the obtained information and geological structure, a comprehensive prediction result can be obtained.

[0125] By combining the advantages of both models, the accuracy and reliability of predictions can be improved. By combining different models, the limitations of a single model can be reduced, and the accuracy of predictions can be improved. Different models may be applicable to different geological conditions and application scenarios. By selecting suitable models to combine, a wider range of predictions and application scenarios can be covered.

[0126] In practical applications, the exponential model is represented as: Where k is the penetration rate, and k_0 is the base penetration rate. Porosity is the porosity, and m is a parameter of the exponential model. The value of m is determined by fitting experimental data or by selecting an empirical formula.

[0127] The elastic modulus and Poisson's ratio of conglomerate reservoirs were calculated using the Biot-Gassmann model. The elastic modulus and Poisson's ratio of conglomerate reservoirs are expressed as follows:

[0128]

[0129] Where E_m and v_m are the elastic modulus and Poisson's ratio of the sandstone reservoir under dry conditions, respectively, and E_w and v_w are the elastic modulus and Poisson's ratio of the sandstone reservoir under pore water saturation conditions, respectively.

[0130] The Poisson's ratio sensitivity index for sandstone and conglomerate reservoirs is calculated using the following formula:

[0131] O = (v_w - v) / (ΔP / P_0), where O is the Poisson's ratio sensitivity index, ΔP is the change in pore water pressure, and P_0 is the atmospheric pressure.

[0132] The permeability sensitivity index for sandstone and conglomerate reservoirs is calculated using the following formula:

[0133] Where I is the penetration rate sensitivity index.

[0134] A regression model was established using Poisson's ratio sensitivity index and permeability sensitivity index as independent variables, and reservoir productivity or reserves as dependent variables.

[0135] By comprehensively considering the influence of factors such as basement fault slope, source supply size, reservoir thickness, sedimentary facies type, effective reservoir sensitive attributes, physical property changes, oil and gas migration distance, sand body transport capacity, and pressure sensitive parameters, the oil-bearing height of each conglomerate fan body was finally quantitatively predicted.

[0136] By establishing a pressure sensitivity model for sandstone and conglomerate reservoirs, we can assess the reservoir characteristics and reserve distribution under various geological conditions, thereby reducing exploration and development risks and avoiding resource waste. We can also conduct in-depth analysis and research on reservoirs to predict their production capacity and pressure response at different development stages, thus providing scientific and reasonable solutions for oil and gas development.

[0137] S7. Through three-dimensional visualization technology, quantitative boundary features are presented to users to enable analysis and comparison of the internal structure and properties of the sector.

[0138] Data can be converted into point cloud data, and then point cloud visualization technology can be used to present the data to users in the form of point clouds. In the presentation of quantitative boundary features, point cloud data is converted into a surface model, and then the boundary features are marked with different colors or textures to highlight their location and morphological characteristics;

[0139] The data contour can also be extracted, and then contour visualization technology can be used to present the data contour to the user in two or three dimensions. In the presentation of quantitative boundary features, the boundary features are marked with different colors or lines to highlight their position and shape features.

[0140] Beneficial effects:

[0141] 1. This invention establishes a phase model of sandstone and conglomerate bodies across the entire region, enabling a more refined division and interpretation of geological history and a better understanding of geological evolution. It helps identify geological structural features at different stages, such as faults and folds, facilitating a comprehensive analysis of regional geological structures. Furthermore, it provides a better understanding of the formation and distribution patterns of sandstone and conglomerate fans within the region, thereby improving the accuracy of sandstone and conglomerate body predictions. Moreover, it offers guidance on the migration and change patterns of groundwater, rocks, and other materials within the region, providing a scientific basis for engineering exploration and data and methodological support for in-depth research in related fields. This is of great significance for regional economic and social development and resource utilization.

[0142] 2. This invention establishes an information transformation neural network to convert data information into image information, directly presenting the characteristics and distribution of the data information. Based on image data, a more effective model is constructed, thereby improving model performance and accuracy. Furthermore, a semi-supervised learning model based on transform is used to identify and analyze the boundaries inside and outside the sector, utilizing unlabeled data to enhance the model's generalization ability, thus improving classification accuracy. Simultaneously, the transform-based method further enhances data diversity and robustness, thereby further improving model performance. A large amount of unlabeled data is used to train the model, thereby reducing labeling costs. This method makes full use of all data, thereby increasing data utilization and improving model performance.

[0143] 3. This invention can accurately characterize the morphology and structural features of sandstone and conglomerate fans, improving the accuracy and depth of research and analysis. Sandstone and conglomerate fans are one of the important types of oil and gas reservoirs. By using a quantitative boundary characterization method based on basement fracture control, the morphology and structural features of sandstone and conglomerate fans can be more accurately characterized, providing more reliable geological models and reservoir prediction results, and providing support and guidance for oil and gas exploration and development.

[0144] 4. This invention establishes a pressure sensitivity model for sandstone and conglomerate reservoirs to assess reservoir characteristics and reserve distribution under various geological conditions, thereby reducing exploration and development risks and avoiding resource waste; it also conducts in-depth analysis and research on reservoirs to predict reservoir productivity and pressure response at different development stages, thus providing a scientific and reasonable solution for oil and gas development.

[0145] The above specific embodiments further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for delineating the boundaries of sandstone and conglomerate fan bodies based on basement fractures, characterized in that, The boundary delineation method includes: Step S1: Collect remote sensing image data and geological survey data of the conglomerate fan, and obtain measurement data of the conglomerate fan; Step S2: Based on the remote sensing image data and geological survey data, establish a phase model of the sandstone and conglomerate bodies in the whole area and determine the preliminary characteristics; Step S3: Based on the basement fracture characteristics, obtain the boundary morphology and location information of the conglomerate fan. Step S4: Optimize and characterize the boundary morphology and location information of the sandstone and conglomerate fan to obtain the quantitative boundary characteristics of the fan; Step S5: Divide the sandstone and conglomerate fan into multiple clusters based on quantitative boundary characteristics, and determine the boundary of each cluster; Step S6: Establish a pressure sensitivity model for sandstone and conglomerate reservoirs to predict the oil-bearing height of each phase of sandstone and conglomerate fan bodies; Step S7: Present the quantitative boundary features to the user.

2. The method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture, as described in claim 1, is characterized in that... Step S2: Based on the remote sensing image data and geological survey data, establish a phase model of the sandstone and conglomerate bodies in the entire area, and determine the preliminary characteristics, specifically including: The well-seismic fine calibration was adopted, combined with actual drilled well core samples, logging and analysis data, and based on the aforementioned remote sensing image data and geological survey data; By using a comparative analysis of well-connected skeleton profiles, a phase model of the sandstone and conglomerate bodies in the entire area was established, and preliminary characteristics were determined.

3. The method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture, as described in claim 2, is characterized in that... Step S2: Based on the remote sensing image data and geological survey data, establish a phase model of the sandstone and conglomerate bodies in the entire area, and determine the preliminary characteristics, specifically including: Determine the variable n to be analyzed i , i = 1, 2, ... a, including groundwater content and the abundance of petroleum mineral resources; Determine multiple levels of multiple factors r t-p t = 1, 2, ..., b, where multiple factors include topography, geological structure, and lithology, and p ∈ {the set of level values ​​corresponding to the factors}. The total variance is SS tatal =SS between +SS within SS tatal This represents the variance of all data. Within-group variance Indicates that r is being analyzed t-p The variance of the set of data, where For the rth t-p Observations of the set of data For the rth t-p The average of the group data; Between-group variance This represents the variance among multiple groups. For the rth t-p Group sample size The average of all data; F-value = MSB / MSW, which represents the ratio of the mean square MSB between groups to the mean square MSW within groups. The larger the F-value, the more significant the difference.

4. The method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture, as described in claim 1, is characterized in that... Step S3: Obtaining the boundary morphology and location information of the conglomerate fan based on the basement fracture characteristics specifically includes: Based on the phase model of the sandstone and conglomerate bodies in the whole region, profiles were extracted for each phase. Fault lines were extracted based on velocity differences; An information transformation neural network, including an input layer, a transformation layer, an optimization layer, and an output layer, is constructed to convert collected data information into image information. At the same time, a semi-supervised learning model based on transform is constructed to identify and analyze the boundaries of conglomerate fan bodies.

5. The method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture, as described in claim 4, is characterized in that... Step S3: Obtaining the boundary morphology and location information of the conglomerate fan based on the basement fracture characteristics further includes: The input layer and transformation layer are fully connected. The transformation layer transforms the data information. The current data information has N features, specifically: sr1=ω 21 Q j +θ,j∈[1,N] Right now Where sr1 represents the input of the transformation layer, ω 21 Q represents the association weight between the input layer and the transformation layer. j Let θ represent the j-th data feature input to the input layer, θ represent the bias of the transformation layer, f represent the activation function of the transformation layer, T represent the bias, K represent the transformation factor, and sc1 represent the output of the transformation layer. f represents the learning factor. z Let z represent the z-th transformation object, and γ represent the stability factor; The transformation layer passes the processing results to the optimization layer for optimization. The calculation process of the optimization layer is as follows: sr2=ω 32 sc1+θ' Where sr2 represents the input to the optimization layer, ω 32 The weights represent the correlation between the transformation layer and the optimization layer, θ' represents the bias of the optimization layer, β represents the control parameter, and ρ represents the preset adjustment parameter; j ≠ j'; f j' This represents the transformation object corresponding to the j'th data information feature.

6. The method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture, as described in claim 1, is characterized in that... Step S4: Optimizing and characterizing the boundary morphology and location information of the conglomerate fan to obtain the quantitative boundary features of the fan specifically includes: The sandstone and conglomerate fan was divided into multiple units of uniform size and shape using gridding or voxelization methods; Calculate the distance and angle between each element and the base fracture, and use the distance and angle between each element and the base fracture as the element's characteristic value.

7. The method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture, as described in claim 1, is characterized in that... Step S5: Dividing the sandstone and conglomerate fan into multiple clusters based on quantitative boundary characteristics, and determining the boundary of each cluster specifically includes: For the boundary of each cluster, a fitting algorithm is used to fit a curve to obtain a model of the cluster boundary; The cluster boundaries were optimized and corrected to obtain the final sandstone and conglomerate fan body boundaries.

8. The method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture, as described in claim 7, is characterized in that... After obtaining the final boundary of the conglomerate fan, the process also includes: performing error analysis on the boundary.

9. The method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture, as described in claim 7, is characterized in that... The error analysis of the boundary specifically includes: boundary shape, boundary change trend, and boundary location.

10. The method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture, as described in claim 1, is characterized in that... Step S6: Establishing a pressure sensitivity model for sandstone and conglomerate reservoirs to predict the oil-bearing height of each phase of the sandstone and conglomerate fan specifically includes: A pressure sensitivity model for sandstone and conglomerate reservoirs was established. Based on information and geological structure, two suitable models were selected and combined to obtain comprehensive prediction results. The exponential model is as follows: Where k is the penetration rate, and k_0 is the base penetration rate. It represents porosity, and m is a parameter of the exponential model; The elastic modulus E and Poisson's ratio v of sandstone and conglomerate reservoirs were calculated based on the Biot-Gassmann model. Poisson's ratio sensitivity index O for sandstone and conglomerate reservoirs; permeability sensitivity index I for sandstone and conglomerate reservoirs; A regression model was established using Poisson's ratio sensitivity index and permeability sensitivity index as independent variables, and reservoir productivity or reserves index as dependent variables.

11. The method for delineating the boundary of a sandstone-conglomerate fan based on a basement fracture, as described in claim 1, is characterized in that... Step S7: Presenting the quantitative boundary features to the user specifically includes: Using 3D visualization technology, quantitative boundary features are presented to users, enabling the analysis and comparison of the internal structure and properties of the sector.

Images