Method and program product for reservoir heterogeneity determination

Abstract

The application discloses a reservoir heterogeneity determination method and a program product, and relates to the field of reservoir heterogeneity determination.The method comprises the following steps: obtaining a target core image corresponding to a core sample of a target reservoir; inputting the target core image into a pore region segmentation network to obtain a binary image; wherein the binary image distinguishes between pore regions and mineral regions included in the target core image; constructing a pore network model based on the binary image; determining a geometric correction coefficient based on the binary image, and determining a geometric heterogeneity component corresponding to the core sample according to the geometric correction coefficient and the pore network model; and determining a topological heterogeneity component corresponding to the core sample based on the pore network model; and determining a heterogeneity coefficient corresponding to the target reservoir according to the geometric heterogeneity component and the topological heterogeneity component; thereby achieving accurate characterization of the heterogeneity of the target reservoir and improving the accuracy and reliability of the determination of the heterogeneity of the target reservoir.

Description

Technical Field

[0001] This invention relates to the field of geological reservoir evaluation technology, and in particular to a method and program product for determining reservoir heterogeneity. Background Technology

[0002] Reservoir heterogeneity is a core geological factor controlling hydrocarbon flow patterns, pressure propagation characteristics, and fracturing effectiveness. Its essence stems from the spatial differences in the geometry, connectivity, and topology of mineral grains, pores, and fractures at different spatial scales. This multi-scale, multi-type non-uniform distribution directly determines the dominant channels, stagnant zones, and stress-sensitive response zones of fluids within the reservoir, making it one of the key control factors restricting the efficient development and refined stimulation of unconventional reservoirs.

[0003] Current technologies for determining reservoir heterogeneity largely rely on physical analysis of core samples, assessing it through macroscopic parameters such as statistical pore size distribution, specific surface area, and porosity. These methods typically acquire data with limited dimensions, reflecting only the average properties of the core sample and failing to comprehensively capture the complex spatial differences in pore structure at the microscale. This limits the accuracy of reservoir heterogeneity prediction. Therefore, a new method for determining reservoir heterogeneity is urgently needed to improve its accuracy and reliability. Summary of the Invention

[0004] This invention provides a method and program product for determining reservoir heterogeneity, which solves the problem that the characterization of reservoir heterogeneity in related technologies mostly relies on the statistical analysis of macroscopic parameters of core samples, resulting in a single data dimension, difficulty in capturing microscopic pore differences, and limited prediction accuracy.

[0005] According to one aspect of the present invention, a method for determining reservoir heterogeneity is provided, the method comprising:

[0006] Obtain the target core image corresponding to the core sample of the target reservoir;

[0007] The target core image is input into a pore region segmentation network to obtain a binarized image; wherein the binarized image distinguishes between the pore region and the mineral region included in the target core image;

[0008] A porosity network model is constructed based on the binarized image;

[0009] Geometric correction coefficients are determined based on the binarized image; and the geometric heterogeneity component corresponding to the core sample is determined based on the geometric correction coefficients and the pore network model.

[0010] The topological heterogeneity components corresponding to the core sample are determined based on the pore network model.

[0011] The heterogeneity coefficient corresponding to the target reservoir is determined based on the geometric heterogeneity component and the topological heterogeneity component.

[0012] According to another aspect of the present invention, a reservoir heterogeneity determination apparatus is provided, the apparatus comprising:

[0013] The image acquisition module is used to acquire the target core image corresponding to the core sample of the target reservoir;

[0014] An image segmentation module is used to input the target core image into a pore region segmentation network to obtain a binarized image; wherein the binarized image distinguishes and displays the pore region and mineral region included in the target core image;

[0015] The model building module is used to build a porosity network model based on the binarized image;

[0016] The first component determination module is used to determine the geometric correction coefficient based on the binarized image, and to determine the geometric heterogeneity component corresponding to the core sample according to the geometric correction coefficient and the pore network model.

[0017] The second component determination module is used to determine the topological heterogeneity component corresponding to the core sample based on the pore network model.

[0018] The heterogeneity determination module is used to determine the heterogeneity coefficient corresponding to the target reservoir based on the geometric heterogeneity component and the topological heterogeneity component.

[0019] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0020] At least one processor; and

[0021] A memory communicatively connected to the at least one processor; wherein,

[0022] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the reservoir heterogeneity determination method according to any embodiment of the present invention.

[0023] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the reservoir heterogeneity determination method according to any embodiment of the present invention.

[0024] According to another aspect of the present invention, embodiments of this disclosure also provide a computer program product, including a computer program that, when executed by a processor, implements the reservoir heterogeneity determination method as described in any of the embodiments of this disclosure.

[0025] The technical solution of this invention includes the following steps: First, acquiring a target core image corresponding to a core sample of the target reservoir; acquiring the target core image reflects the true microstructure of the target reservoir, laying the foundation for subsequent analysis; next, inputting the target core image into a pore region segmentation network to obtain a binarized image; wherein, the binarized image distinguishes between the pore region and mineral region included in the target core image; automated region segmentation is achieved based on the pore region segmentation network, significantly improving region identification efficiency and segmentation accuracy; next, constructing a pore network model based on the binarized image; transforming static two-dimensional image information into a three-dimensional pore network model with topological connections to reflect the true spatial distribution of pores inside the core sample; finally, determining geometric correction coefficients based on the binarized image, according to the... The geometric correction coefficient and the pore network model are used to determine the geometric heterogeneity component corresponding to the core sample. The geometric correction coefficient effectively corrects calculation errors and improves the accuracy of geometric heterogeneity component calculation. Next, the topological heterogeneity component corresponding to the core sample is determined based on the pore network model. Quantitative analysis is performed based on topological structures such as pore connectivity and pore node arrangement to accurately capture deep structural characteristics such as pore connectivity quality and network density. Finally, the heterogeneity coefficient corresponding to the target reservoir is determined based on the geometric and topological heterogeneity components. The fusion of dual-dimensional components avoids the one-sidedness of single-dimensional evaluation, making the heterogeneity coefficient more closely match the actual geological characteristics of the reservoir, accurately depicting the differences in the reservoir's microstructure, and achieving a comparable, standardized, and quantitative expression of reservoir microstructural heterogeneity.

[0026] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0028] Figure 1 This is a flowchart of a method for determining reservoir heterogeneity according to Embodiment 1 of the present invention;

[0029] Figure 2 This is a flowchart of a method for determining reservoir heterogeneity according to Embodiment 2 of the present invention;

[0030] Figure 3 This is a schematic diagram of a core sample obtained by focused ion beam scanning electron microscopy in the depth direction according to a method for determining reservoir heterogeneity provided in Embodiment 2 of the present invention.

[0031] Figure 4 This is a schematic diagram of a binarized image of a core sample used in a method for determining reservoir heterogeneity according to Embodiment 2 of the present invention.

[0032] Figure 5 This is a schematic diagram of the probability density distribution of topological characteristic parameters in a reservoir heterogeneity determination method provided in Embodiment 2 of the present invention.

[0033] Figure 6 This is a schematic diagram of the initial geometric heterogeneity component verification of a reservoir heterogeneity determination method provided in Embodiment 2 of the present invention;

[0034] Figure 7 This is a schematic diagram of a regular cubic network in a reservoir heterogeneity determination method according to Embodiment 2 of the present invention;

[0035] Figure 8 This is a schematic diagram of a series planar radial network for a reservoir heterogeneity determination method provided in Embodiment 2 of the present invention;

[0036] Figure 9 This is a schematic diagram of a three-dimensional radial network for a method of determining reservoir heterogeneity according to Embodiment 2 of the present invention;

[0037] Figure 10 This is a schematic diagram of a core sample network for a reservoir heterogeneity determination method provided in Embodiment 2 of the present invention;

[0038] Figure 11 This is a schematic diagram of a reservoir heterogeneity determination device according to Embodiment 3 of the present invention;

[0039] Figure 12 This is a schematic diagram of the structure of an electronic device that implements the reservoir heterogeneity determination method of the present invention. Detailed Implementation

[0040] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0041] It should be noted that the terms "first," "second," "initial," "target," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0042] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0043] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0044] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0045] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0046] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0047] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0048] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0049] Example 1

[0050] Figure 1 This document provides a flowchart of a method for determining reservoir heterogeneity according to Embodiment 1 of the present invention. This embodiment is applicable to situations where reservoir heterogeneity is evaluated. The method can be executed by a reservoir heterogeneity determination device, which can be implemented in hardware and / or software, optionally through electronic devices such as mobile terminals, PCs, or servers. Figure 1 As shown, the method may specifically include:

[0051] S110. Obtain the target core image corresponding to the core sample of the target reservoir.

[0052] In this embodiment of the invention, the target reservoir can be understood as a reservoir region whose heterogeneity needs to be determined. The core sample can be a rock sample collected from the target reservoir that reflects the original rock characteristics of the reservoir. The target core image can be an image captured by an imaging device and preprocessed to clearly show the internal structure of the core sample.

[0053] Based on the above scheme, optionally, obtaining the target core image corresponding to the core sample of the target reservoir includes: obtaining an initial core image corresponding to the core sample of the target reservoir based on focused ion beam scanning electron microscopy; processing the initial core image based on a nonlocal mean filtering algorithm to obtain a second image; performing Fourier transform processing on the second image to obtain a third image; performing band-stop filtering processing on the third image to obtain a fourth image; performing inverse Fourier transform processing on the fourth image to obtain a fifth image; performing grayscale normalization processing on the fifth image to obtain a sixth image; and performing depth stacking restoration on the sixth image to obtain the target core image.

[0054] The Focused Ion Beam Scanning Electron Microscope (FIB-SEM) is a device for high-resolution scanning imaging of core samples, capable of accurately capturing the internal microstructure of the core and thus obtaining an initial core image. The initial core image can be an unprocessed image of the core sample directly captured by the FIB-SEM, which may contain noise, insufficient resolution, and uneven brightness. The second image can be a core image with noise suppressed after processing using a nonlocal mean filtering algorithm. The third image can be a frequency domain image obtained after Fourier transform processing, clearly showing the frequency distribution characteristics of the image and facilitating the distinction between effective and noise signals. The fourth image is a frequency domain image obtained after band-stop filtering, precisely suppressing noise components within a specific frequency range. The fifth image is a spatial domain image obtained after inverse Fourier transform processing, with noise effectively removed and significantly improved microstructural clarity. The sixth image can be a uniformly grayscale image obtained after grayscale normalization. The target core image is a three-dimensional microstructure image obtained after depth stacking and reconstruction.

[0055] Specifically, a continuous sequence of interface microscopic images can be obtained using FIB-SEM to serve as the initial core images corresponding to the core samples of the target reservoir. For example, the surface of the core sample from the target reservoir can be finely polished using mechanical polishing and ion beam (FIB) to prepare a smooth and clean core sample slice. Then, the core sample slice is placed in a field emission scanning electron microscope (FESEM) and backscattered electron imaging is performed under an accelerating voltage of 5 kV to obtain a continuous cross-sectional image sequence with a pixel resolution of 512 × 512 and a spatial resolution of 10 nm / pixel. The continuous cross-sectional image sequence comprises 512 frames, which can completely cover the specified depth range of the core sample.

[0056] Due to unavoidable interference factors such as electron beam drift and mechanical vibration during the acquisition process, the aforementioned continuous two-dimensional images may exhibit spatial deviations, resulting in inter-frame drift. Without correction, this would severely affect the accuracy of subsequent depth stacking reconstruction. Therefore, a normalized cross-correlation (NCC) algorithm can be used to calculate the two-dimensional cross-correlation matrix for each adjacent frame, extracting the displacement peak coordinates as the inter-frame offset. Spatial translation compensation is then applied to each frame to eliminate the inter-frame displacement distortion introduced by electron beam thermal drift and mechanical vibration, resulting in an aligned continuous cross-sectional image sequence. Aligned 2D slices can be stacked along the Z-axis depth direction.

[0057] Considering the high complexity and computational cost of directly processing 3D images, a 2D slice approach can be used. Frame-by-frame optimization is performed, followed by depth-direction stacking and reconstruction to obtain a complete target core image. Optionally, this can be based on an aligned sequence of consecutive cross-sectional images. A two-dimensional initial core image is determined. Then, to address issues such as electron beam shot noise and background brightness drift in the initial core image, a non-local means filter (NLM) can be used for background noise removal. By weighted averaging of similar pixel blocks in the image, the non-local similarity of the image is fully considered, achieving image smoothing while denoising, effectively avoiding the detail loss problem caused by traditional filtering algorithms. For example, the pixel values ​​after non-local means filtering can be determined based on the following method:

[0058] ;

[0059] in, Represents pixels The new pixel value after filtering; Represents pixels The original pixel value, pixel The nearest pixels within the search window; Represented in pixels The search window is centered on a radius of 10. The normalization coefficient is such that the sum of all weights equals 1. , Represents pixels For pixels The influence weight. In an example of this invention, the spatial standard deviation can be preset. Strength standard deviation The search window radius is 10, and the local neighborhood radius is... The above parameter combination has been verified by historical data. It can effectively suppress electron beam noise and background drift while maintaining clear aperture boundaries, thus achieving noise reduction and preserving image details.

[0060] Processing with a nonlocal mean filtering algorithm can significantly reduce omnidirectional high-frequency signals in the spectral image and remove high-frequency noise. Nonlocal mean filtering has a good noise reduction effect, minimal impact on image smoothing, and effectively preserves the detailed features of the core microstructure. The second image is obtained after nonlocal mean filtering. Next, to address the potential local periodic background noise in the second image, Fourier transform combined with band-stop filtering can be used to further improve image quality.

[0061] Fourier transform converts spatial domain signals into frequency domain signals, transforming the grayscale information of an image into a spectrum. In a spectrum image, faster grayscale changes correspond to higher frequencies, and slower changes correspond to lower frequencies. Stripes and artifacts in core images often appear as long, straight stripes along a single direction, which manifest as bright lines in the corresponding direction in the spectrum image. Frequency domain filtering can precisely remove the frequency signals in that direction from the spectrum image, effectively removing stripes and artifacts and improving image clarity.

[0062] For the two-dimensional second image, a Fourier transform can be performed based on the following formula to obtain the third image:

[0063] ;

[0064] Next, the third image can be band-stop filtered based on the following formula to obtain the fourth image:

[0065] ;

[0066] Finally, after filtering, the fourth image can be subjected to inverse Fourier transform based on the following formula to obtain the fifth image. :

[0067] ;

[0068] in, Represents the pixels in the second image Pixel values ​​in the spatial domain; For pixels Complex values ​​in the frequency domain after Fourier transform and These are frequency coordinates, corresponding to the frequency components in the horizontal and vertical directions, respectively; and The number of rows and columns of the image, i.e., the number of rows and columns of the image. and Dimensions in the direction; Numerical basis functions are used to project pixel values ​​from the spatial domain to the frequency domain. The imaginary unit; For the filter at the frequency point The response value at the location; Representing frequency point Distance to the origin of the frequency domain; Indicates the first The center frequency that needs to be suppressed is the location of the detected noise dominant frequency; The parameter used to control the filter bandwidth can be set to 10 in this embodiment of the invention. This parameter can be flexibly adjusted according to the actual noise situation to ensure the noise suppression effect.

[0069] Next, the fifth image can be targeted. Gray-level normalization is performed to improve the gray-level contrast between pores and mineral phases, eliminate the influence of brightness differences in different regions, and enhance the processing consistency and generalization ability of multi-source data, providing a unified image basis for subsequent pore region segmentation. Gray-level normalization can be performed based on the following formula:

[0070] ;

[0071] in, Represents the number of pixels in the image after normalization. Pixel values; For the pixels in the fifth image The original pixel values; This represents the minimum grayscale value of the image. The maximum grayscale value of the image is denoted as 1. By normalizing the grayscale values, the normalized pixel values ​​can be distributed in the range of [0, 255], thus obtaining the sixth image.

[0072] Finally, the sixth image can be deep-stacked and restored to obtain the target core image. Here, z can be used to label the layer number or depth of the 2D image within the entire image sequence to ensure the accuracy of the 3D reconstruction. Simultaneously, the target core image can be saved as a TIFF file. TIFF format supports high-resolution image storage, fully preserving the microscopic details of the image, facilitating subsequent processing and analysis. The target core image obtained after the above preprocessing exhibits high consistency in brightness uniformity, noise control level, and structural clarity, providing a standardized and reproducible data foundation for subsequent pore region segmentation and model input, effectively improving the accuracy and reliability of subsequent heterogeneity analysis.

[0073] To address the systematic errors unique to FIB-SEM imaging of shale reservoirs, such as electron beam drift, charging effect, and curtain artifacts, a cascaded preprocessing workflow was established, incorporating normalized cross-correlation algorithms, Fast Fourier Transform (FFT), and nonlocal mean filtering. This workflow is adapted to the characteristics of focused ion beam scanning electron microscopy (FIB-SEM) imaging data, effectively solving the problem that existing image processing workflows developed for computed tomography (CT) are unsuitable for FIB-SEM data and produce poor processing results. This significantly improves the processing quality and efficiency of core images.

[0074] S120. Input the target core image into a pore region segmentation network to obtain a binarized image; wherein the binarized image distinguishes between the pore region and the mineral region included in the target core image.

[0075] The pore region segmentation network can be used as a model to distinguish different regions in a core image, thereby separating pore regions from mineral regions. A binarized image can be understood as an image processed by the pore region segmentation network, containing only two grayscale values: one corresponding to pore regions and the other to mineral regions, visually representing the distribution of these two types of regions. A pore region can be understood as a void-like area within the core sample that is not filled with minerals, where reservoir fluids such as oil, gas, and water can be stored and flow. A mineral region can be understood as a solid area within the core sample composed of rock minerals, filling the pores.

[0076] After image preprocessing, a pore region segmentation network can be used to automatically identify and segment the porous and mineral phases. Specifically, this pore region segmentation network can employ a depth segmentation network based on an improved U-NET structure. An adaptive threshold histogram suppression module is introduced on top of the standard U-Net structure to address the problem of blurred pore boundaries caused by local exposure differences and background drift in Scanning Electron Microscope (SEM) images. This further improves the segmentation accuracy of pore and mineral regions, enabling regional segmentation of complex reservoirs with intricate pore structures and blurred boundaries.

[0077] Based on the above scheme, optionally, the pore region segmentation network includes a block histogram equalization module and a pore region segmentation module. The step of inputting the target core image into the pore region segmentation network to obtain a binarized image includes: performing histogram equalization processing on the target core image based on the block histogram equalization module to obtain a first image; and performing backscattered electron signal intensity detection on the first image based on the pore region segmentation module to label the pore region and the mineral region on the first image respectively, thereby obtaining the binarized image.

[0078] The block histogram equalization module divides the target core image into multiple small blocks, performs histogram equalization on each block separately, improves image contrast, alleviates local brightness unevenness, and makes the boundaries between pore and mineral regions clearer. The first image can be the image obtained after processing by the block histogram equalization module, with improved contrast and clearer pore boundaries. The pore region segmentation module can be used to analyze the first image, distinguish between pore and mineral regions, and annotate them.

[0079] Specifically, a normalized high-resolution grayscale image can be used. As input data, the grayscale signal can be used to reflect the difference in backscattered electron (BSE) intensity. Mineral regions have higher BSE signal intensity, corresponding to higher grayscale values, while pore regions have lower BSE signal intensity, corresponding to lower grayscale values. Firstly, the target core image can be processed using a block-based histogram equalization module. An adaptive histogram block equalization mechanism is introduced to divide the image into multiple overlapping sub-blocks. Histogram equalization is performed independently on each sub-block to obtain a locally enhanced brightness image. ;in, ; For pixels in the processed image Pixel values; For the pixels in the input image Pixel values; Indicates the first in the image The average brightness level of each sub-block; Indicates the first in the image Standard deviation of each sub-block; The target is the global standard deviation; The target is the global mean. and Determined based on global grayscale statistics. By using block equalization processing, interference from local brightness differences in SEM images can be effectively suppressed, improving the separability of aperture edges.

[0080] Optionally, the aperture region segmentation network can employ a symmetric encoder-decoder framework, wherein the encoder consists of multiple layers of convolution and max pooling, and each convolution layer can be represented as: ;in, For the first Layer feature map; ReLU activation function is used to introduce non-linear features and enhance the expressive power of the network; The convolution kernel and bias are used to extract local features of the image; The derivation represents a convolution operation; the decoder gradually restores the spatial resolution through deconvolution and skip connections, fusing the deep features extracted by the encoder with the shallow features to ensure the accuracy of the segmentation results. Feature fusion can be represented as: ,in, This indicates an upsampling operation used to restore the spatial resolution of an image; Concat indicates feature concatenation, used to fuse features from different layers of the encoder. Indicates the first in the decoder path Feature map of the layer; For the fusion of the first Layer feature map.

[0081] Optionally, multi-scale convolutional modules (1×1, 3×3, and 5×5 convolutions in parallel) can be added to the low-level feature pathways to enhance the detection capability of fine pores. During network training, a composite loss function can be used. , Network training effectively solves the problem of imbalance between foreground (pore region) and background (mineral region) in image segmentation; among which, This represents the total loss function value; This represents the balancing weight coefficient, used to adjust the proportion of cross-entropy loss and Dice loss. It can be set to 0.5, a value that has been verified by historical data to achieve the optimal balance between the two losses. Cross-entropy loss is used to measure the difference between the predicted probability distribution and the true label. The Dice loss is used to address the foreground-background imbalance problem in image segmentation. The model's prediction results; This is a real label; The intersection of the predicted result and the true label, i.e., the number of correctly predicted pixels; The total number of pixels predicted as foreground; This represents the total number of foreground pixels in the true label. Training samples may include manually labeled aperture regions to ensure the reliability of the training data. The optimizer can be an Adaptive Moment Estimation (Adam) optimizer, and the initial learning rate can be set to 1×10⁻⁶. −4 The batch size can be set to 8, and the training can be iterated until the model converges to ensure that the model has good segmentation accuracy.

[0082] During the inference process, the backscattered electron signal intensity of the first image can be detected through the pore region segmentation module, and output pixel by pixel in the coordinate system. Probability distribution of the pixel belonging to the aperture category The pore region and the mineral region are labeled on the first image to obtain the binarized image. Pore binarization is implemented based on an adaptive threshold. , This represents the result after pore binarization, and a binarized image is obtained based on the pore binarization result. 100 is an empirical threshold parameter that can be calibrated and adjusted according to actual scenario requirements to ensure the accuracy of the binarized structure and achieve precise differentiation between porous and mineral regions.

[0083] S130. Construct a pore network model based on the binarized image.

[0084] Among them, the pore network model can be a model constructed based on a binary image to simulate the spatial distribution and connection relationship of pores and throats inside the rock core. The pore network model uses pores as nodes and throats as edges, and can accurately quantify the connectivity characteristics of the pore structure.

[0085] Based on the above scheme, optionally, the step of constructing a pore network model based on the binarized image includes: performing Euclidean distance transformation on the binarized image to obtain a distance transform volume; performing three-dimensional maximum extraction on the distance transform volume, and determining the local maximum points as pore center nodes to construct a pore network model; wherein, in the pore network model, pores are nodes and throats are edges.

[0086] The distance transform volume is a 3D data volume obtained through Euclidean distance transformation, where each voxel value represents the distance from that voxel to the nearest mineral region. A local maximum (Local Maximum) is a pixel in the distance transform volume whose value is greater than all its surrounding pixels; this point corresponds to the center of the pore and can be used as the pore center node in the pore network model. The pore center node is the node representing the pore in the pore network model.

[0087] Specifically, based on obtaining a binarized image, multiple binarized images can be used... By performing stacking and reconstruction along the Z-axis, a complete binary mask image of the pores is obtained. The pore binary mask image clearly shows the distribution range of pores and minerals in the three-dimensional space of the core sample; then, for Perform Euclidean distance transformation to obtain the distance transformation volume. Each voxel value represents the distance from that voxel to the nearest solid boundary. Furthermore, a three-dimensional maximum extraction is performed on the DEDT, identifying local maxima as pore center nodes. Connecting channels between adjacent nodes are used as throats to construct a pore network model with pores as nodes and throats as edges. This is used to record the spatial coordinates of each pore node and the connection relationships between pore nodes.

[0088] For example, such as Figure 10 As shown, for a shale reservoir core sample, the above method can identify 648 pore nodes and 1644 throat connections. The constructed pore network model can accurately reflect the pore connectivity characteristics of the core sample, providing reliable support for subsequent heterogeneity analysis.

[0089] S140. Determine the geometric correction coefficient based on the binarized image, and determine the geometric heterogeneity component corresponding to the core sample according to the geometric correction coefficient and the pore network model.

[0090] The geometric correction coefficient can be a correction parameter calculated based on the pore features of the binarized image. It is used to correct the geometric deviation of the pore network model and improve the accuracy of subsequent heterogeneity calculations. The geometric heterogeneity component can be a quantitative indicator reflecting the inhomogeneity of the reservoir pore structure in terms of geometric morphology, size distribution, etc.

[0091] Based on the above scheme, optionally, determining the geometric correction coefficient based on the binarized image includes: determining the number of pore pixels in the pore region of the binarized image and the total number of pixels in the binarized image; determining the porosity based on the ratio of the number of pore pixels to the total number of pixels in the image; and performing linear fitting on the binarized image based on box counting to obtain the pore fractal dimension; and determining the geometric correction coefficient based on the porosity and the pore fractal dimension.

[0092] The number of pore pixels can be the total number of pixels representing pore regions in a binarized image, indirectly reflecting the size and distribution range of the pore regions. The total number of pixels in the image can be the total number of all pixels in the binarized image. Porosity can be used to quantitatively characterize the proportion of pore space to the total volume in a core sample. The pore fractal dimension can be a quantitative indicator used to characterize the geometric complexity of the pore region; the larger the fractal dimension, the more complex and irregular the pore structure.

[0093] Specifically, porosity can be determined based on the following formula. :

[0094] ;

[0095] in, This indicates the number of pore pixels in the pore region; This represents the total number of pixels in the binarized image.

[0096] Furthermore, this can be achieved through fractal dimension. This represents the geometric complexity of the pores. For example, the fractal dimension can be calculated based on the box-counting method. Placed in a size of Count the number of boxes covering the pores in the grid. The system records multiple sets of statistical values ​​as the box size changes, and calculates the fractal dimension based on the following formula:

[0097] ;

[0098] During the calculation process, multiple different scales can be selected. By analyzing multiple groups and A linear fit is used to determine an approximate value of the fractal dimension, which can represent the complexity of the pore boundary at different scales and reflect the multi-scale spatial distribution characteristics of the porous system.

[0099] Furthermore, a geometric correction coefficient can be determined based on the porosity and the fractal dimension of the pores. For example, the average value of the porosity and fractal dimension obtained from core statistics of a preset region can be acquired, such as... Assuming the porosity calculated above... The fractal dimension is 0.1516. The value is 2.6362, and then the ratio of each to the average is calculated. , Finally based on and Calculate the geometric correction factor , Furthermore, the machine learning model can be pre-trained based on the porosity and fractal dimension of multiple core samples from a preset region to construct a geometric correction coefficient prediction model. This allows for the rapid determination of the geometric correction coefficient directly based on the calculated porosity and pore fractal dimension, thereby improving the efficiency of coefficient calculation.

[0100] Based on the above scheme, optionally, determining the geometric heterogeneity component corresponding to the core sample according to the geometric correction coefficient and the pore network model includes: extracting parameters from the pore network model based on topological feature parameters of multiple feature dimensions to obtain the mean and standard deviation corresponding to multiple topological feature parameters; determining the coefficient of variation of the topological feature parameters according to the mean and standard deviation of the topological feature parameters; determining the first weight corresponding to each of the multiple topological feature parameters according to the coefficient of variation of the multiple topological feature parameters; and determining the geometric heterogeneity component according to the coefficient of variation of the topological feature parameters, the first weight, and the geometric correction coefficient.

[0101] Among them, the feature dimensions of the topological feature parameters can be different angles used to describe the topological characteristics of the porous network model. Each feature dimension can correspond to at least one topological feature parameter. Multiple feature dimensions of topological feature parameters can be used to comprehensively characterize the structural properties of the porous network, avoiding the one-sidedness caused by a single parameter. Topological feature parameters can be parameters used to quantify information such as the shape, size, and connectivity of porous network topologies like pores and throats. The mean can be the arithmetic mean of multiple values ​​of the same topological feature parameter, reflecting the overall average level of that topological feature parameter. The standard deviation can be the degree of deviation of multiple values ​​of the same topological feature parameter from the mean, reflecting the discrete distribution of that topological feature parameter. The coefficient of variation can be understood as the ratio of the standard deviation to the mean, eliminating the influence of parameter magnitude and quantifying the degree of parameter dispersion. The first weight can be a weight value determined based on the coefficient of variation of each topological feature parameter, used to measure the degree of influence of different topological feature parameters on the geometric non-homogeneity component.

[0102] Specifically, parameters can be extracted from multiple pore nodes and throat edges in the pore network model based on topological feature parameters of multiple feature dimensions, obtaining multiple sets of values ​​for each topological feature parameter, and then calculating the mean value corresponding to each of the multiple topological feature parameters. and standard deviation Next, based on the first The mean and standard deviation of the first topological feature parameter determine the second... coefficient of variation of each topological feature parameter , Furthermore, based on the coefficients of variation of the multiple topological feature parameters, a first weight corresponding to each topological feature parameter is determined through normalization processing. , .

[0103] Based on the above scheme, optionally, in order to comprehensively and accurately characterize the geometric features of the porous network, the feature dimension corresponding to the topological feature parameter includes at least one of the following: the equivalent sphere radius of the pores. Pore ​​coordination number Radius of the equivalent circle of the throat Length of the larynx Pore ​​shape factor Laryngeal shape factor The pore equivalent sphere radius can be understood as the radius of a sphere when the pores in a core sample are approximated as spheres. It can be directly read from the maximum value of the distance transformation and is used to quantify pore size, reflecting the pore size. The pore coordination number can be understood as the number of other pores connected to each pore node through the throat edge. It can be determined based on the number of throats directly connected to that pore and is used to quantify pore connectivity. The higher the coordination number, the better the pore connectivity. The throat equivalent circle radius can be understood as the radius of the circle when the throat is approximated as a circle. It can be determined based on the radius of the largest inscribed circle of the throat cross-section and is used to quantify throat size. The larger the throat equivalent circle radius, the stronger the throat's flow conductivity. The throat length can be understood as the actual length of the throat connecting two pores. It can be determined based on the center-to-center distance between two adjacent pores and is used to quantify the throat's spatial size. The shorter the throat length, the lower the resistance to fluid flow. The pore shape factor can be understood as a topological characteristic parameter used to quantify the irregularity of pore geometry. It can be calculated based on the ratio of pore volume to surface area; the larger the shape factor, the more irregular the pore shape. Similarly, the throat shape factor can be understood as a topological characteristic parameter used to quantify the irregularity of throat geometry. It can be determined based on the ratio of throat cross-sectional area to the square of its perimeter; the larger the shape factor, the more irregular the throat shape.

[0104] For example, such as Figure 5 As shown in Table 1, parameters were extracted from the pore network model of a core sample, yielding the topological characteristic parameters, and the corresponding coefficients of variation were calculated. :

[0105]

[0106] Table 1

[0107] Furthermore, the first weight is calculated based on the aforementioned coefficient of variation. The first weight corresponding to each topological feature parameter is obtained through normalization. Simultaneously, the individual values ​​of each topological feature parameter are calculated, as shown in Table 2.

[0108]

[0109] Table 2

[0110] Finally, the geometrical heterogeneity component can be determined based on the coefficient of variation and first weight of the topological feature parameters, and the geometrical correction coefficient, according to the following formula. To achieve independent quantitative assessment of the heterogeneity of geometric parameters:

[0111] ;

[0112] To further demonstrate that the coefficients of variation determined by the above normalization calculation method can accurately characterize the joint heterogeneity of the six types of topological feature parameters, a Monte Carlo simulation can be performed with random values ​​of [0.1-1] for each coefficient of variation. Furthermore, the initial geometric heterogeneity component under the maximum coefficient of variation among the six topological feature parameters with a fixed mean coefficient of variation can be solved. The result is as follows Figure 6 As shown, this indicates that whether it is with the increase of the mean of the coefficient of variation or with the increase of the dispersion of the coefficient of variation itself, All of them increased accordingly.

[0113] S150. Determine the topological heterogeneity component corresponding to the core sample based on the pore network model.

[0114] Among them, the topological heterogeneity component can be a quantitative indicator used to reflect the non-uniformity of the reservoir pore network in terms of topological structure, such as node connection mode and connectivity.

[0115] Optionally, analysis can be performed based on a pore network model. By extracting the topological connectivity features of the pore network, the heterogeneity of the spatial topological connectivity structure of the pore network can be quantitatively analyzed, the topological heterogeneity components corresponding to the core sample can be determined, the discrete features of the pore network topology can be accurately captured, and the one-sidedness of heterogeneity assessment caused by focusing only on geometric features can be avoided, thus ensuring the comprehensiveness and accuracy of the assessment results.

[0116] S160. Determine the heterogeneity coefficient corresponding to the target reservoir based on the geometric heterogeneity component and the topological heterogeneity component.

[0117] The heterogeneity coefficient can be a quantitative value that comprehensively characterizes the degree of heterogeneity of the target reservoir by combining geometric heterogeneity components and topological heterogeneity components through weighted fusion and other methods.

[0118] Based on the above scheme, optionally, determining the heterogeneity coefficient corresponding to the target reservoir according to the geometric heterogeneity component and the topological heterogeneity component includes: calculating the correlation between the geometric heterogeneity component and the reservoir heterogeneity to determine a second weight corresponding to the geometric heterogeneity component; calculating the correlation between the topological heterogeneity component and the reservoir heterogeneity to determine a third weight corresponding to the topological heterogeneity component; and performing weighted fusion based on the geometric heterogeneity component and its corresponding second weight, and the topological heterogeneity component and its corresponding third weight to obtain the heterogeneity coefficient corresponding to the target reservoir.

[0119] The second weight measures the influence of geometric heterogeneity components on heterogeneity coefficients. The third weight measures the influence of topological heterogeneity components on heterogeneity coefficients.

[0120] In determining geometrically non-homogeneous components and topological inhomogeneous components Based on this, the correlation coefficient between the geometric heterogeneity component and reservoir heterogeneity can be calculated through correlation analysis, thereby determining the second weight corresponding to the geometric heterogeneity component; similarly, the correlation coefficient between the topological heterogeneity component and reservoir heterogeneity can be calculated to determine the third weight corresponding to the topological heterogeneity component, where the second weight... With the third weight The sum is 1. Weighted fusion is performed based on the geometric heterogeneity component and its corresponding second weight, and the topological heterogeneity component and its corresponding third weight, to obtain the heterogeneity coefficient corresponding to the target reservoir. ,Right now .

[0121] For example, suppose , If correlation analysis determines that both the second and third weights are 0.5, then... This heterogeneity coefficient comprehensively reflects the degree of heterogeneity of the core sample in two dimensions: geometric morphology and spatial topological connectivity. From the contributions of the two components, This indicates that the pore throat geometric parameters of the core sample are highly discrete in their internal distribution, with large coefficients of variation for parameters such as pore radius and coordination number, and strong geometric heterogeneity. This indicates that the energy distribution of the pore network connectivity pathways in the core sample exhibits a certain degree of inhomogeneity. While dominant seepage channels are developed, they have not yet formed an extremely concentrated single main channel, and the topological heterogeneity is at a moderate level. After weighted fusion of the two data points... The overall heterogeneity level is moderate, and the contribution of the geometric dimension is significantly greater than that of the topological dimension. This reflects that the heterogeneity of the reservoir core sample is mainly driven by the uneven distribution of pore throat size, while the concentration effect of topological connectivity paths is relatively weak and has a secondary impact on the overall heterogeneity.

[0122] The second and third weights can be flexibly adjusted according to the actual needs of the scenario. For example, when the research objective focuses on the evaluation of seepage paths or the analysis of the connectivity of fracture networks, the third weight can be appropriately increased. To enhance the influence of the topological dimension and more accurately reflect the impact of topological structure on reservoir permeability; when the research objective focuses on characterizing pore throat geometric parameters or rock physical modeling, the topological dimension can be appropriately reduced. This enhances the influence of geometric dimensions, more accurately reflecting the impact of pore-throat geometry on reservoir storage and permeability. Furthermore, the heterogeneity coefficient... The high, medium, and low threshold values ​​can be determined independently based on the actual reservoir type and operational needs of the target reservoir, in order to adapt to the heterogeneity assessment needs of different basins and different types of reservoirs, and improve the universality and applicability of the reservoir heterogeneity determination method.

[0123] The technical solution of this invention includes the following steps: First, acquiring a target core image corresponding to a core sample of the target reservoir; acquiring the target core image reflects the true microstructure of the target reservoir, laying the foundation for subsequent analysis; next, inputting the target core image into a pore region segmentation network to obtain a binarized image; wherein, the binarized image distinguishes between the pore region and mineral region included in the target core image; automated region segmentation is achieved based on the pore region segmentation network, significantly improving region identification efficiency and segmentation accuracy; next, constructing a pore network model based on the binarized image; transforming static two-dimensional image information into a three-dimensional pore network model with topological connections to reflect the true spatial distribution of pores inside the core sample; finally, determining geometric correction coefficients based on the binarized image, according to the... The geometric correction coefficient and the pore network model are used to determine the geometric heterogeneity component corresponding to the core sample. The geometric correction coefficient effectively corrects calculation errors and improves the accuracy of geometric heterogeneity component calculation. Next, the topological heterogeneity component corresponding to the core sample is determined based on the pore network model. Quantitative analysis is performed based on topological structures such as pore connectivity and pore node arrangement to accurately capture deep structural characteristics such as pore connectivity quality and network density. Finally, the heterogeneity coefficient corresponding to the target reservoir is determined based on the geometric and topological heterogeneity components. The fusion of dual-dimensional components avoids the one-sidedness of single-dimensional evaluation, making the heterogeneity coefficient more closely match the actual geological characteristics of the reservoir, accurately depicting the differences in the reservoir's microstructure, and achieving a comparable, standardized, and quantitative expression of reservoir microstructural heterogeneity.

[0124] Example 2

[0125] Figure 2 This is a flowchart illustrating a method for determining reservoir heterogeneity according to Embodiment 2 of the present invention, further describing specific implementation methods for determining topological heterogeneity components. Specific implementation methods can be found in the description of this embodiment. Technical features that are the same as or similar to those in the foregoing embodiments will not be repeated here. Figure 2 As shown, the method may specifically include:

[0126] S210. Obtain the target core image corresponding to the core sample of the target reservoir.

[0127] S220. Input the target core image into a pore region segmentation network to obtain a binarized image; wherein the binarized image distinguishes between the pore region and the mineral region included in the target core image.

[0128] S230. Construct a pore network model based on the binarized image.

[0129] S240. Determine the geometric correction coefficient based on the binarized image, and determine the geometric heterogeneity component corresponding to the core sample according to the geometric correction coefficient and the pore network model.

[0130] S250. Determine the throat conduction coefficients corresponding to multiple throats based on the topological characteristic parameters corresponding to multiple throats in the pore network model.

[0131] In a pore network model, a throat is a channel connecting two pore nodes. As a critical path for fluid flow in the reservoir, the shape and size of the throat directly affect the fluid flow capacity. The throat conductivity coefficient is a quantitative indicator used to characterize the throat's ability to conduct fluid; the larger the conductivity coefficient, the stronger the fluid transport capacity of the throat.

[0132] Optionally, the throat conductivity coefficients corresponding to multiple throats can be determined based on the topological characteristic parameters corresponding to multiple throats in the pore network model. Specifically, the throat conductivity coefficients can be derived based on the Hagen-Poiseuille Equation (HP Equation) under the circular pipe cross-section approximation. This accurately describes the laminar flow law of fluid in a circular pipe, reflecting the volumetric flow rate of fluid passing through the throat under a unit pressure difference, and precisely quantifying the throat's conductivity. The throat conductivity coefficients can be calculated based on the following formula:

[0133] ;

[0134] in, Pore ​​nodes and pore nodes The equivalent circle radius of the throat passage; The throat length is denoted by . The throat conductivity coefficient is dominated by the fourth power of the throat radius and is highly sensitive to changes in throat geometry. It can effectively distinguish the differences in geometric characteristics of different throats, and thus accurately reflect the differences in the contribution of different throats to fluid transport capacity.

[0135] S260. Construct a weighted graph Laplace matrix based on the throat conductivity coefficients corresponding to the multiple throats and the pore network model.

[0136] Among them, the weighted graph Laplacian matrix can be a graph theory matrix constructed based on the porosity network model, with the throat conduction coefficient as the weight. It can be used to quantitatively characterize the connection relationship between nodes and edges and the weight distribution in the porosity network.

[0137] Optionally, the larynx diversion coefficients corresponding to multiple larynxes can be... As edge weights, a weighted adjacency matrix A is constructed. Elements in the matrix are defined as those of the pore nodes. and pore nodes When there is a larynx connection When there is no connection For example, assuming the total number of pore nodes is 648 and the total number of throat connections is 1645, then the adjacency matrix A is a real symmetric sparse matrix of 648×648.

[0138] Furthermore, a weighted degree matrix D can be constructed, which is a diagonal matrix, and its 6th degree matrix is... The diagonal elements are defined as pore nodes. The sum of the larynx's diversion coefficients of all connecting larynxes, i.e. The weighting degree of a node reflects its overall connectivity in the weighted network. A higher weighting degree indicates a stronger fluid transport capacity through the throat connection. Finally, the weighted graph Laplace matrix L can be obtained by subtracting the adjacency matrix A from the weighting degree matrix D, i.e., L = DA. The weighted graph Laplace matrix L is an n×n real symmetric positive semi-definite matrix, where n is the total number of pore nodes. The sum of the elements in each row is always 0. Diagonal elements equal the weighting degree of the corresponding pore node, and off-diagonal non-zero elements equal the negative values ​​of the corresponding throat conductivity coefficients. The matrix structure of L fully encodes the global topological connectivity of the pore network and the flow weights of each connecting channel, organically combining topological structure and conductivity, laying the foundation for subsequent spectral analysis.

[0139] S270. Perform eigenvalue decomposition on the weighted graph Laplacian matrix to determine the sequence of topological eigenvalues.

[0140] The topological eigenvalue sequence can be a series of eigenvalues ​​arranged in a specific order, obtained by eigenvalue decomposition of the weighted graph Laplacian matrix. Each eigenvalue corresponds to a connectivity mode of the porous network, and the magnitude and distribution of the eigenvalues ​​can accurately reflect the uniformity and connectivity characteristics of the network topology.

[0141] Optionally, the weighted graph Laplacian matrix can be subjected to eigenvalue decomposition to obtain a sequence of topological eigenvalues ​​arranged in ascending order. Since L is a positive semi-definite matrix, according to the mathematical properties of positive semi-definite matrices, all eigenvalues ​​are non-negative, and therefore the first eigenvalue... The first eigenvalue is always equal to 0, and its corresponding eigenvector represents a uniform distribution of values ​​across all network nodes. This eigenvalue only represents the overall network translation pattern and does not carry any information about topological differences. Therefore, it is discarded in subsequent calculations of topological heterogeneity to avoid interfering with the calculation results. The second eigenvalue... Known as the Fiedler Value or algebraic connectivity, its corresponding eigenvector can "most naturally" divide the network into two parts in a topological sense. The magnitude of the Fiedler Value is closely related to the heterogeneity of the network topology. A smaller value indicates a lower cost for the partition, meaning the network is more prone to connectivity breaks at some point, resulting in stronger topological heterogeneity. Conversely, a larger Fiedler value indicates better network connectivity and a more homogeneous topology. For example, if... It can be equal to 1.8649. This value indicates that the network as a whole is not easy to break, there is no obvious single bottleneck, and the overall seepage will not be interrupted due to the blockage of a certain throat. This indicates that the topology of the pore network is relatively homogeneous and the connectivity is stable. to Corresponding to higher-order connectivity patterns of the network at different spatial scales, the larger the eigenvalue, the more refined and localized the connectivity pattern, reflecting the differences in connectivity characteristics in local regions of the porous network. The complete sequence of topological eigenvalues ​​constitutes the spectral representation of the network topology, encompassing connectivity information at all scales from global to local. Compared to traditional single topological parameters, this sequence can more comprehensively and multidimensionally characterize the topological features of porous networks, providing rich data support for the accurate quantification of subsequent topological heterogeneity.

[0142] S280. Calculate the probability distribution of eigenvalues ​​based on the eigenvalue sequence of the topological structure to determine the spectral entropy.

[0143] Among them, spectral entropy can be used to quantify the degree of disorder in the distribution of topological eigenvalues. The larger the spectral entropy, the more uneven the topological structure of the pore network, the more dispersed the distribution of connectivity modes, and the stronger the concentration of fluid flow paths.

[0144] Optionally, to ensure the scientific validity and rationality of spectral entropy calculation, the eigenvalue sequence should first be transformed into a probability distribution, and then the spectral entropy should be calculated based on the probability distribution. Specifically, all non-zero eigenvalues ​​can be normalized to a probability distribution using the following formula, giving it probabilistic significance, eliminating the influence of the absolute value of the eigenvalues, and ensuring the comparability of spectral entropies for porous networks of different sizes:

[0145] ;

[0146] Next, the spectral entropy of the eigenvalue probability distribution can be calculated based on the following formula. This measures the degree of disorder in the distribution of eigenvalues, and thus reflects the heterogeneity of the porosity of the network topology.

[0147] ;

[0148] By transforming the distribution characteristics of topological feature values ​​into quantifiable entropy values, the limitations of traditional topological analysis, which can only provide qualitative descriptions, are avoided, and quantitative characterization of topological heterogeneity is achieved. At the same time, through normalization processing, the influence of network size on spectral entropy calculation results is effectively eliminated, and the comparability of topological heterogeneity among different reservoir core samples is improved.

[0149] S290. Normalize the spectral entropy and the theoretical maximum value of the spectral entropy to determine the topological heterogeneity component; determine the heterogeneity coefficient corresponding to the target reservoir based on the geometric heterogeneity component and the topological heterogeneity component.

[0150] The theoretical maximum value of spectral entropy can be understood as the maximum value that spectral entropy can reach under specific conditions. It can be used to normalize the calculated spectral entropy and eliminate the influence of dimensions.

[0151] Spectral entropy It can be used to measure the uniformity of the probability distribution of eigenvalues, that is, the balance of energy distribution across connectivity modes at various scales in a porous network, and its magnitude is positively correlated with topological heterogeneity. Specifically, when all non-zero eigenvalues ​​are completely equal, It exhibits a uniform distribution, and the spectral entropy reaches its theoretical maximum value. The first type corresponds to an ideal void network with a completely homogeneous topology; when the eigenvalues ​​are highly concentrated in a few modes, the spectral entropy approaches 0, corresponding to a degenerate network with an extremely heterogeneous topology. In this case, the fluid mainly flows along a few dominant paths, and the topological inhomogeneity is extremely strong. To eliminate network size... The influence of the absolute value of spectral entropy on the comparability of core samples of different sizes can be assessed by considering the theoretical maximum value of the spectral entropy. Normalize.

[0152] Specifically, the topological inhomogeneity components can be determined by normalizing the spectral entropy and its theoretical maximum value using the following formula:

[0153] ;

[0154] Assuming the calculated This indicates that the energy distribution uniformity of the pore network connectivity in the core sample reaches 66.7% of the theoretical maximum value, meaning that connectivity is non-uniformly distributed. Some pore throat channels contribute significantly more to conductivity than the average level, and fluid tends to concentrate its flow along a few dominant paths. The network topological heterogeneity is at a moderate level. Furthermore, if we combine this with the above... The results further indicate that the overall connectivity of the pore network is good, with no structural bottlenecks that could disrupt global connectivity at a single point. The network is not prone to overall flow interruption due to local throat failure. The two indicators mentioned above can be interpreted together to reveal the typical topological characteristics of the reservoir core sample: a complete connectivity framework and limited path diversity. During water injection development, fluids tend to rapidly penetrate along dominant throats, leading to low matrix porosity utilization and significant room for improvement in reservoir sweep efficiency. By jointly interpreting spectral entropy and Fiedler values, both the overall topological homogeneity and local connectivity vulnerability of the network can be simultaneously captured, providing richer structural information for a more refined evaluation of reservoir heterogeneity than a single statistical measure.

[0155] For example, to verify the ability of the spectral entropy index to quantify and distinguish different topological structures and to ensure the reliability of the calculation of topological heterogeneity components, three 5×5×5 control networks can be constructed. Each network has 125 nodes, and the geometric parameters such as node spatial location, pore radius, and shape factor are completely identical, differing only in topological connectivity, to eliminate the interference of geometric parameters on topological analysis and ensure the accuracy of the verification results. The specific control network design is as follows: Figure 7 , Figure 8 , Figure 9 As shown, grid 1 is a regular cube connection, with each internal node evenly connected to 6 face neighbors, totaling 300 connections, representing an ideal homogeneous topology baseline; grid 2 is a hub-and-chain hybrid structure, with 5 hub nodes each responsible for approximately 24 ordinary nodes, and the hubs are linked together, totaling 124 connections, representing a moderately heterogeneous topology; grid 3 is a single-hub star structure, with the central node directly connected to all 124 other nodes, totaling 124 connections, representing an extremely heterogeneous topology, such as... Figure 10 As shown, assuming the network connection number of the core sample is 1644, The spectral entropy calculation results of the three networks and the core sample network can be summarized in Table 3:

[0156]

[0157] Table 3

[0158] According to the table above, we can see that... The topological heterogeneity indices of the three control networks monotonically increased with increasing network topological heterogeneity, exhibiting a clear gradient distribution and significant differences. (Core samples) The core sample's pore network exhibits a clear concentration trend, falling between a hub-and-chain hybrid and a single star-shaped pattern, and more closely resembling a single star. This indicates that the connectivity pathways of the pore network show a clear concentration trend, with a few large pore throats dominating the main fluid transport channels. This characteristic aligns with the typical topological features of tight reservoir pore networks. The aforementioned comparative experiments effectively validate the spectral entropy index's ability to quantify the spatial topological heterogeneity of pore networks, further demonstrating the scientific validity and reliability of the topological heterogeneity component calculations.

[0159] The technical solution of this invention firstly determines the throat conductivity coefficients corresponding to multiple throats based on the topological feature parameters corresponding to multiple throats in the pore network model; by performing parameter conversion based on the actual pore throat topological properties, the differences in conductivity of different throats can be accurately distinguished, achieving refined quantification of microscopic seepage capacity and providing basic parameters for subsequent matrix construction; nextly, a weighted graph Laplace matrix is ​​constructed based on the throat conductivity coefficients and the pore network model; the matrix constructed based on the throat conductivity coefficients closely matches the actual connectivity of the pore network, accurately restoring the overall architecture of the fluid flow network inside the core sample; then, the topological structure feature value sequence is determined by performing eigenvalue decomposition on the weighted graph Laplace matrix; and finally, the matrix is ​​used to construct the matrix. Feature decomposition enables in-depth deconstruction of network topology, efficiently extracting hidden deep structural patterns in pore networks and transforming complex spatial connectivity relationships into ordered numerical sequences, simplifying the quantitative analysis of complex topologies. Next, spectral entropy is determined by calculating the probability distribution of eigenvalues ​​based on the topological feature value sequence. Quantifying the degree of disorder in the pore topology through spectral entropy directly reflects the chaotic arrangement and connectivity differences of the pore network, achieving a quantitative characterization of topological disorder. Finally, topological heterogeneity components are determined by normalizing the spectral entropy and its theoretical maximum value. Normalization eliminates data interference caused by sample scale and model size, enabling lateral comparability between different core samples and achieving standardized quantification of topological heterogeneity.

[0160] Example 3

[0161] Figure 11 This is a schematic diagram of a reservoir heterogeneity determination device provided in Embodiment 3 of the present invention. This device is used to execute the reservoir heterogeneity determination method provided in any of the above embodiments. This device and the reservoir heterogeneity determination methods of the above embodiments belong to the same inventive concept. Details not described in detail in the embodiments of the reservoir heterogeneity determination device can be referred to the embodiments of the above reservoir heterogeneity determination methods. Figure 11 As shown, the device includes: an image acquisition module 310, an image segmentation module 320, a model construction module 330, a first component determination module 340, a second component determination module 350, and a heterogeneity determination module 360.

[0162] The system includes: an image acquisition module 310 for acquiring a target core image corresponding to a core sample of the target reservoir; an image segmentation module 320 for inputting the target core image into a pore region segmentation network to obtain a binarized image, wherein the binarized image distinguishes between pore regions and mineral regions included in the target core image; a model construction module 330 for constructing a pore network model based on the binarized image; a first component determination module 340 for determining a geometric correction coefficient based on the binarized image, and determining a geometric heterogeneity component corresponding to the core sample based on the geometric correction coefficient and the pore network model; a second component determination module 350 for determining a topological heterogeneity component corresponding to the core sample based on the pore network model; and a heterogeneity determination module 360 ​​for determining a heterogeneity coefficient corresponding to the target reservoir based on the geometric heterogeneity component and the topological heterogeneity component.

[0163] The technical solution of this invention includes the following steps: First, the image acquisition module 310 acquires a target core image corresponding to the core sample of the target reservoir. Acquiring the target core image reflects the true microstructure of the target reservoir, laying the foundation for subsequent analysis. Next, the image segmentation module 320 inputs the target core image into a pore region segmentation network to obtain a binarized image. The binarized image distinguishes between pore regions and mineral regions included in the target core image. Automated region segmentation is achieved based on the pore region segmentation network, significantly improving region identification efficiency and segmentation accuracy. Then, the model construction module 330 constructs a pore network model based on the binarized image. This transforms static two-dimensional image information into a three-dimensional pore network model with topological connections, reflecting the true spatial distribution of pores within the core sample. Finally, the first component determination module 340 determines the geometry based on the binarized image. The geometrical correction coefficient is used to determine the geometric heterogeneity component corresponding to the core sample based on the geometrical correction coefficient and the pore network model. This effectively corrects calculation errors and improves the accuracy of geometric heterogeneity component calculations. Next, the second component determination module 350 determines the topological heterogeneity component corresponding to the core sample based on the pore network model. Quantitative analysis is performed based on topological structures such as pore connectivity and pore node arrangement to accurately capture deep structural features such as pore connectivity quality and network density. Finally, the heterogeneity determination module 360 ​​determines the heterogeneity coefficient corresponding to the target reservoir based on the geometrical and topological heterogeneity components. This fusion of dual-dimensional components avoids the bias of single-dimensional evaluation, making the heterogeneity coefficient more closely match the actual geological characteristics of the reservoir, accurately characterizing the differences in the reservoir's microstructure, and achieving a comparable, standardized, and quantitative expression of the reservoir's microstructural heterogeneity.

[0164] Based on the above scheme, optionally, the first component determination module 340 includes a topology feature parameter calculation submodule, a variation parameter determination submodule, a first weight determination submodule, and a first component determination submodule. Specifically, the topology feature parameter calculation submodule is used to extract parameters from the porosity network model based on topology feature parameters of multiple feature dimensions to obtain the mean and standard deviation corresponding to multiple topology feature parameters; the variation parameter determination submodule is used to determine the variation coefficient of the topology feature parameters based on the mean and standard deviation of the topology feature parameters; the first weight determination submodule is used to determine the first weight corresponding to each topology feature parameter based on the variation coefficient of the multiple topology feature parameters; and the first component determination submodule is used to determine the geometric non-homogeneity component based on the variation coefficient of the topology feature parameters, the first weight, and the geometric correction coefficient.

[0165] Based on the above scheme, optionally, the second component determination module 350 includes: a throat conduction coefficient determination submodule, a matrix construction submodule, a sequence construction submodule, a spectral entropy determination submodule, and a second component determination submodule. Specifically, the throat conduction coefficient determination submodule is used to determine the throat conduction coefficients corresponding to multiple throats based on the topological feature parameters corresponding to multiple throats in the pore network model; the matrix construction submodule is used to construct a weighted graph Laplacian matrix based on the throat conduction coefficients corresponding to the multiple throats and the pore network model; the sequence construction submodule is used to perform eigenvalue decomposition on the weighted graph Laplacian matrix to determine the topological eigenvalue sequence; the spectral entropy determination submodule is used to calculate the eigenvalue probability distribution based on the topological eigenvalue sequence to determine the spectral entropy; and the second component determination submodule is used to perform normalization processing based on the spectral entropy and the theoretical maximum value of the spectral entropy to determine the topological heterogeneity component.

[0166] Based on the above scheme, optionally, the heterogeneity determination module 360 ​​includes: a second weight determination submodule, a third weight determination submodule, and a heterogeneity determination submodule. The second weight determination submodule is used to determine the second weight corresponding to the geometric heterogeneity component by performing correlation calculation between the geometric heterogeneity component and the reservoir heterogeneity; the third weight determination submodule is used to determine the third weight corresponding to the topological heterogeneity component by performing correlation calculation between the topological heterogeneity component and the reservoir heterogeneity; the heterogeneity determination submodule is used to perform weighted fusion of the geometric heterogeneity component and its corresponding second weight, and the topological heterogeneity component and its corresponding third weight, to obtain the heterogeneity coefficient corresponding to the target reservoir.

[0167] Based on the above scheme, optionally, the first component determination module 340 includes: a porosity determination submodule, a pore fractal dimension determination submodule, and a geometric correction coefficient determination submodule. The porosity determination submodule is used to determine the number of pore pixels in the pore region of the binarized image and the total number of pixels in the binarized image, and to determine the porosity based on the ratio of the number of pore pixels to the total number of pixels. The pore fractal dimension determination submodule is used to perform linear fitting on the binarized image based on box counting to obtain the pore fractal dimension. The geometric correction coefficient determination submodule is used to determine geometric correction coefficients based on the porosity and the pore fractal dimension.

[0168] Based on the above scheme, optionally, the model construction module 330 includes: a distance transform volume construction module and a model construction submodule. The distance transform volume construction module is used to perform Euclidean distance transformation based on the binarized image to obtain a distance transform volume; the model construction submodule is used to extract three-dimensional maxima based on the distance transform volume, determining the local maxima points as pore center nodes to construct a pore network model; wherein, in the pore network model, pores are nodes and throats are edges.

[0169] Based on the above scheme, optionally, the pore region segmentation network includes a block histogram equalization module and a pore region segmentation module, and the image segmentation module 320 includes a first image processing module and an image segmentation submodule. The first image processing module is used to perform histogram equalization processing on the target core image based on the block histogram equalization module to obtain a first image; the image segmentation submodule is used to perform backscattered electron signal intensity detection on the first image based on the pore region segmentation module to label the pore region and the mineral region on the first image respectively, to obtain the binarized image.

[0170] Based on the above scheme, optionally, the image acquisition module 310 includes: an initial three-dimensional core image acquisition submodule, a second image acquisition submodule, a third image acquisition submodule, a fourth image acquisition submodule, a fifth image acquisition submodule, and a target core image acquisition submodule. The initial core image acquisition submodule is used to acquire an initial core image corresponding to the core sample of the target reservoir based on focused ion beam scanning electron microscopy; the second image acquisition submodule is used to process the initial core image based on a nonlocal mean filtering algorithm to obtain a second image; the third image acquisition submodule is used to perform Fourier transform processing on the second image to obtain a third image; the fourth image acquisition submodule is used to perform band-stop filtering processing on the third image to obtain a fourth image; the fifth image acquisition submodule is used to perform inverse Fourier transform processing on the fourth image to obtain a fifth image; the sixth image acquisition submodule is used to perform grayscale normalization processing on the fifth image to obtain a sixth image; and the target core image acquisition submodule is used to perform depth stacking reconstruction on the sixth image to obtain the target core image.

[0171] Based on the above scheme, optionally, the feature dimension corresponding to the topological feature parameter includes at least one of the following: pore equivalent sphere radius, pore coordination number, throat equivalent circle radius, throat length, pore shape factor, and throat shape factor.

[0172] The reservoir heterogeneity determination apparatus provided in this embodiment of the invention can execute the reservoir heterogeneity determination method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0173] Example 4

[0174] Figure 12 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0175] like Figure 12As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0176] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0177] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as methods for determining reservoir heterogeneity.

[0178] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication unit 19, or installed from storage unit 18, or installed from ROM 12. When the computer program is executed by processor 11, it performs the functions defined in the methods of the embodiments of the present invention.

[0179] In some embodiments, the reservoir heterogeneity determination method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the reservoir heterogeneity determination method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the reservoir heterogeneity determination method by any other suitable means (e.g., by means of firmware).

[0180] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0181] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0182] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0183] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0184] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0185] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0186] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0187] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for determining reservoir heterogeneity, characterized in that, include: Obtain the target core image corresponding to the core sample of the target reservoir; The target core image is input into a pore region segmentation network to obtain a binarized image; wherein the binarized image distinguishes between the pore region and the mineral region included in the target core image; A porosity network model is constructed based on the binarized image; Geometric correction coefficients are determined based on the binarized image; and the geometric heterogeneity component corresponding to the core sample is determined based on the geometric correction coefficients and the pore network model. The topological heterogeneity components corresponding to the core sample are determined based on the pore network model. The heterogeneity coefficient corresponding to the target reservoir is determined based on the geometric heterogeneity component and the topological heterogeneity component; The step of determining the geometric correction coefficients based on the binarized image includes: Determine the number of pore pixels in the pore region of the binarized image and the total number of pixels in the binarized image; determine the porosity based on the ratio of the number of pore pixels to the total number of pixels; and, The binarized image is linearly fitted based on box counting to obtain the pore fractal dimension. The geometric correction coefficients are determined based on the porosity and the fractal dimension of the pores.

2. The method for determining reservoir heterogeneity according to claim 1, characterized in that, The step of determining the geometric heterogeneity component corresponding to the core sample based on the geometric correction coefficient and the pore network model includes: The porosity network model is parameter extracted based on topological feature parameters of multiple feature dimensions to obtain the mean and standard deviation of the multiple topological feature parameters; The coefficient of variation of the topological feature parameters is determined based on the mean and standard deviation of the topological feature parameters; The first weights corresponding to the topological feature parameters are determined based on the coefficients of variation of the multiple topological feature parameters. The geometric inhomogeneity component is determined based on the coefficient of variation and first weight of the topological feature parameters, as well as the geometric correction coefficient.

3. The method for determining reservoir heterogeneity according to claim 1, characterized in that, The determination of the topological heterogeneity component corresponding to the core sample based on the pore network model includes: The throat conduction coefficients corresponding to multiple throats are determined based on the topological feature parameters corresponding to multiple throats in the pore network model. A weighted graph Laplace matrix is ​​constructed based on the throat conduction coefficients corresponding to the multiple throats and the pore network model. The weighted graph Laplacian matrix is ​​subjected to eigenvalue decomposition to determine the sequence of topological eigenvalues; The spectral entropy is determined by calculating the probability distribution of eigenvalues ​​based on the eigenvalue sequence of the topological structure. The topological inhomogeneity components are determined by normalizing the spectral entropy and its theoretical maximum value.

4. The method for determining reservoir heterogeneity according to claim 1, characterized in that, The step of determining the heterogeneity coefficient corresponding to the target reservoir based on the geometric heterogeneity component and the topological heterogeneity component includes: The second weight corresponding to the geometric heterogeneity component is determined by calculating the correlation between the geometric heterogeneity component and the reservoir heterogeneity. The third weight corresponding to the topological heterogeneity component is determined by calculating the correlation between the topological heterogeneity component and the reservoir heterogeneity. The heterogeneity coefficient corresponding to the target reservoir is obtained by weighted fusion of the geometric heterogeneity component and its corresponding second weight, the topological heterogeneity component and its corresponding third weight.

5. The method for determining reservoir heterogeneity according to claim 1, characterized in that, The construction of the porosity network model based on the binarized image includes: Based on the binarized image, a Euclidean distance transform is performed to obtain the distance transform volume; Based on the distance transform, a three-dimensional maximum value is extracted, and the local maximum value point is determined as the pore center node to construct a pore network model; wherein, in the pore network model, pores are nodes and throats are edges.

6. The method for determining reservoir heterogeneity according to claim 1, characterized in that, The pore region segmentation network includes a block histogram equalization module and a pore region segmentation module. The step of inputting the target core image into the pore region segmentation network to obtain a binarized image includes: The target core image is subjected to histogram equalization processing based on the block histogram equalization module to obtain a first image; The backscattered electron signal intensity is detected on the first image based on the pore region segmentation module, so as to mark the pore region and the mineral region on the first image respectively, and obtain the binarized image.

7. The method for determining reservoir heterogeneity according to claim 1, characterized in that, The acquisition of the target core image corresponding to the core sample of the target reservoir includes: Initial core images of core samples from the target reservoir were obtained using focused ion beam scanning electron microscopy. The initial core image is processed using a nonlocal mean filtering algorithm to obtain a second image; Perform Fourier transform on the second image to obtain the third image; The third image is subjected to band-stop filtering to obtain the fourth image; Perform an inverse Fourier transform on the fourth image to obtain the fifth image; The fifth image is subjected to grayscale normalization to obtain the sixth image; The sixth image is then subjected to depth stacking reconstruction to obtain the target core image.

8. The method for determining reservoir heterogeneity according to claim 2, characterized in that, The feature dimensions corresponding to the topological feature parameters include at least one of the following: pore equivalent sphere radius, pore coordination number, throat equivalent circle radius, throat length, pore shape factor, and throat shape factor.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the reservoir heterogeneity determination method as described in any one of claims 1-8.

Images