Nanometer-precision millimeter-scale digital core reconstruction method, device, equipment and medium

Abstract

The application discloses a nanometer-precision millimeter-scale digital core reconstruction method, device, equipment and medium, and relates to the technical field of oil and gas field development. The method adopts a denoising diffusion probability model to respectively perform sample expansion on digital cores with different resolutions, including micron, submicron and nanometer resolutions; a pSp super-resolution network is trained by using digital core data sets after expansion of two adjacent resolutions, to obtain a super-resolution reconstruction model with different resolutions; according to a real micron-resolution digital core, the super-resolution reconstruction model with different resolutions is used to perform super-resolution reconstruction on the real micron-resolution digital core step by step, to obtain a reconstructed nanometer-resolution millimeter-scale digital core; and the reconstructed digital core is corrected according to physical parameters of a real core, to obtain a nanometer-precision millimeter-scale digital core. The application can solve the contradiction between resolution and field of view in digital core reconstruction.

Description

Technical Field

[0001] This application relates to the field of oil and gas field development technology, and in particular to a method, apparatus, equipment and medium for digital core reconstruction at the nanometer-precision millimeter scale. Background Technology

[0002] Shale reservoirs exhibit abundant nanoscale pores and micrometer-scale natural fractures, resulting in a multi-scale pore system. Accurate characterization of shale pore and fracture structures is crucial for assessing reservoir permeability and guiding the large-scale, efficient development of shale oil. Currently, the most effective tool for three-dimensional characterization of shale nanoscale pores is the focused ion beam scanning electron microscope (FIB-SEM), with an imaging precision of up to 10 nm, capable of accurately characterizing shale nanopores. Computed tomography (CT) imaging precision ranges from hundreds of nanometers to tens of micrometers, effectively detecting micrometer-scale macropores and microfractures. However, these observation methods suffer from a contradiction between resolution and field of view. High-resolution imaging equipment (such as FIB-SEM) typically has a field of view within 10 μm, making it difficult to simultaneously capture micrometer-scale fracture networks; conversely, large-field-of-view CT scans are limited by resolution thresholds, failing to clearly identify submicrometer and nanopore structures. This inverted relationship between resolution and observation scale makes it difficult for currently constructed shale digital cores to fully characterize the micro- and nano-scale pore and fracture network structure of shale. Summary of the Invention

[0003] The purpose of this application is to provide a method, apparatus, device, and medium for digital core reconstruction at the nanometer-precision millimeter scale, which can solve the contradiction between resolution and field of view in digital core reconstruction, so that the constructed shale digital core can reach the millimeter scale, meet the requirements for accurate characterization of micrometer-level macropores and microcracks, and at the same time, the precision can reach the nanometer level, thereby accurately characterizing the nanometer-level pores in shale.

[0004] To achieve the above objectives, this application provides the following solution.

[0005] In a first aspect, this application provides a method for reconstructing digital cores at the nanometer-precision millimeter scale, comprising: acquiring the physical property parameters of real cores; acquiring digital cores at different resolutions; the different resolutions include micrometer resolution, submicrometer resolution, and nanometer resolution; the micrometer-resolution digital cores are at the millimeter scale; using denoising diffusion probabilistic models (DDPM) to augment the digital cores at different resolutions to obtain digital core datasets augmented at each resolution; training a pSp (pixel2style2pixel) super-resolution network using the digital core datasets augmented at adjacent resolutions to obtain super-resolution reconstruction models at different resolutions; using the super-resolution reconstruction models at different resolutions to perform super-resolution reconstruction of the real micrometer-resolution digital cores step by step to obtain reconstructed nanometer-resolution millimeter-scale digital cores; and correcting the reconstructed nanometer-resolution millimeter-scale digital cores according to the physical property parameters of the real cores to obtain the final nanometer-precision millimeter-scale digital cores.

[0006] Optionally, the physical properties of the actual rock core are obtained, specifically including:

[0007] The physical properties include porosity, permeability, and pore size distribution curves;

[0008] Core porosity was measured using the helium expansion method.

[0009] The permeability of real rock cores was measured using the pulse attenuation method.

[0010] The core pore size distribution curve was measured using nuclear magnetic resonance.

[0011] Optionally, digital core samples at different resolutions can be acquired, specifically including:

[0012] Real rock cores were scanned using a micron-CT device to obtain digital rock cores with micron resolution and millimeter scale. The digital rock cores with micron resolution and millimeter scale were then denoised to obtain denoised digital rock cores with micron resolution and millimeter scale.

[0013] A real rock core was scanned using a nano-CT device to obtain a digital rock core with submicron resolution and micron scale. The submicron resolution and micron scale digital rock core was then denoised to obtain a denoised submicron resolution and micron scale digital rock core.

[0014] A digital core at nanometer resolution and micrometer scale was obtained by scanning a real rock core using a focused ion beam scanning electron microscope. The digital core at nanometer resolution and micrometer scale was then denoised and aligned to obtain a processed digital core at nanometer resolution and micrometer scale.

[0015] Optionally, a denoising diffusion probability model is used to augment digital core samples at different resolutions to obtain digital core datasets augmented for each resolution, specifically including:

[0016] A separate denoising diffusion probability model was trained using digital cores of each resolution to obtain the trained denoising diffusion probability models corresponding to different resolutions.

[0017] Based on the digital cores at each resolution, multiple digital cores are generated using the pre-trained denoising diffusion probability model corresponding to each resolution, and these multiple digital cores are combined to form a digital core dataset expanded for each resolution.

[0018] Optionally, a pSp super-resolution network is trained using digital core datasets augmented with two adjacent resolutions to obtain super-resolution reconstruction models at different resolutions, specifically including:

[0019] Define any two adjacent resolutions in different resolutions as a resolution pair;

[0020] Using the augmented digital cores corresponding to the low and medium resolutions of each resolution pair as input and the augmented digital cores corresponding to the medium and high resolutions of each resolution pair as output, a pSp super-resolution network is trained to obtain the super-resolution reconstruction model for each resolution pair.

[0021] The super-resolution reconstruction models for all resolution pairs are then combined to form super-resolution reconstruction models with different resolutions.

[0022] Optionally, using super-resolution reconstruction models of different resolutions, the real micron-resolution digital core is reconstructed step by step to obtain a reconstructed nanometer-resolution millimeter-scale digital core, specifically including:

[0023] Super-resolution reconstruction models with different resolutions were determined, including submicron resolution super-resolution reconstruction models and nanometer resolution super-resolution reconstruction models;

[0024] Divide the digital core sample with true micron resolution and millimeter scale into n 3 A set of core samples of the same size; n is a positive integer;

[0025] Based on the core samples divided at micrometer resolution, a super-resolution reconstruction model at submicrometer resolution is used to obtain n. 3 A reconstructed digital core with submicron resolution;

[0026] All the reconstructed submicron resolution digital cores were combined according to the division order of core samples at micron resolution to construct a submicron resolution millimeter-scale digital core.

[0027] The reconstructed submicron resolution millimeter-scale digital core was divided into m... 3 A set of core samples of the same size; m is a positive integer;

[0028] Based on the core samples divided at submicron resolution, a super-resolution reconstruction model at nanometer resolution was used to obtain m 3 A reconstructed digital core at nanometer resolution;

[0029] All the reconstructed nanometer-resolution digital cores were combined according to the sub-micrometer resolution core sample division order to construct a nanometer-resolution millimeter-scale digital core.

[0030] Optionally, based on the physical properties of the actual core, the reconstructed nanometer-resolution millimeter-scale digital core is corrected to obtain the final nanometer-precision millimeter-scale digital core, specifically including:

[0031] Determine the porosity threshold based on the porosity of the actual rock core;

[0032] The reconstructed nanometer-resolution millimeter-scale digital core is divided into its skeleton and pores according to the pore threshold, and a three-dimensional pore network model of the core is constructed.

[0033] Based on the three-dimensional pore network model of the rock core, the pore size distribution is calculated;

[0034] If the error between the simulated pore size distribution and the actual pore size distribution of the core is less than a preset threshold, then the permeability of the three-dimensional pore network model of the core is calculated.

[0035] If the error between the calculated permeability and the permeability of the actual core is less than a preset threshold, the reconstructed nanometer-resolution millimeter-scale digital core will be determined as the final nanometer-precision millimeter-scale digital core.

[0036] If the error between the simulated pore size distribution and the actual pore size distribution of the core is greater than or equal to a preset threshold, then the pixels at the edges of the pores and skeleton in the reconstructed nanometer-resolution millimeter-scale digital core are modified, and the process returns to the step of dividing the skeleton and pores of the reconstructed nanometer-resolution millimeter-scale digital core according to the pore threshold to construct a three-dimensional pore network model of the core.

[0037] Secondly, this application provides a nanometer-precision millimeter-scale digital core reconstruction device, which is used to implement the nanometer-precision millimeter-scale digital core reconstruction method described above. The nanometer-precision millimeter-scale digital core reconstruction device includes: a parameter acquisition module, a digital core acquisition module, a digital core sample expansion module, a training module, a reconstruction module, and a correction module.

[0038] The system comprises the following modules: a parameter acquisition module for acquiring the physical properties of real core samples; a digital core acquisition module for acquiring real digital core samples at different resolutions, including micrometer, submicrometer, and nanometer resolutions; and a micrometer resolution digital core at the millimeter scale. A digital core sample expansion module is used to expand the digital core samples at different resolutions using a denoising diffusion probability model, obtaining expanded digital core datasets for each resolution. A training module trains a pSp super-resolution network using the expanded digital core datasets at two adjacent resolutions, obtaining super-resolution reconstruction models at different resolutions. A reconstruction module uses these super-resolution reconstruction models to progressively reconstruct the real micrometer resolution digital cores, obtaining reconstructed nanometer resolution millimeter-scale digital cores. A correction module corrects the reconstructed nanometer resolution millimeter-scale digital cores based on the physical properties of the real core samples, obtaining the final nanometer precision millimeter-scale digital cores.

[0039] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described nanometer-precision millimeter-scale digital core reconstruction method.

[0040] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described nanometer-precision millimeter-scale digital core reconstruction method.

[0041] According to the specific embodiments provided in this application, this application has the following technical effects.

[0042] This application provides a method, apparatus, device, and medium for digital core reconstruction at the nanometer-precision millimeter scale. DDPM is used to augment digital core datasets at different resolutions. Adjacent resolution digital core data pairs from the augmented datasets are used to train a pSp super-resolution network, constructing a multi-level super-resolution reconstruction model. This model enables cross-scale super-resolution reconstruction of digital cores from the micrometer to sub-micrometer to nanometer scales. Finally, the constructed digital cores are corrected to obtain nanometer-precision millimeter-scale digital cores, resolving the conflict between resolution and field of view during digital core reconstruction. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a schematic flowchart of a nanometer-precision millimeter-scale digital core reconstruction method provided in an embodiment of this application.

[0045] Figure 2 This is a schematic diagram of multi-resolution imaging provided in an embodiment of this application.

[0046] Figure 3 This is a schematic diagram of the training process of the denoising diffusion probability model provided in the embodiments of this application.

[0047] Figure 4 A digital core result diagram generated by the denoising diffusion probability model provided in this application embodiment.

[0048] Figure 5 This is a schematic diagram of the pSp super-resolution network structure provided in an embodiment of this application.

[0049] Figure 6 This is a diagram illustrating the reconstruction results of the pSp super-resolution network provided in this embodiment of the application.

[0050] Figure 7 A schematic diagram of the multi-scale digital core reconstruction process provided in this application embodiment.

[0051] Figure 8 This is a schematic diagram of the cross-scale digital core correction process provided in the embodiments of this application.

[0052] Figure 9 A schematic diagram of the functional modules of a nanometer-precision millimeter-scale digital core reconstruction device provided in this application embodiment.

[0053] Figure 10 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0054] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0055] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0056] Thanks to the powerful fitting ability of neural networks, deep learning methods have been widely used for super-resolution reconstruction of low-resolution shale images. Super-resolution methods establish mapping relationships between images of different scales, reconstructing multi-scale digital cores. This can clearly present the topological morphology of small-scale pores and accurately depict the large-scale micro-fracture structure of shale. Chinese patent CN113609696A discloses a multi-scale digital core construction method based on image fusion. This method directly adds two-dimensional pores extracted from two-dimensional large-scale moiré stitching (MAPS) to CT digital cores, creating false pore connectivity and ineffective seepage channels, making it difficult to directly reflect connectivity and topological relationships in three-dimensional space. Chinese patent CN105487121A discloses a method for constructing multi-scale digital cores by combining CT scan images and electrophysiological imaging images. This method uses information such as porosity distribution and pore size distribution in electrophysiological imaging images to construct large-scale planar images, and fuses CT scan images to construct large-scale three-dimensional digital cores in the longitudinal direction. However, the digital cores constructed by this method have low accuracy and cannot reflect the nanoscale pore structure information of tight reservoirs such as shale. Chinese patent CN116012545A discloses a multi-scale digital core modeling method. This method first constructs a low-resolution digital core using CT experiments, and then combines prior information on nanopores obtained by SEM to generate nanopores on the original low-resolution digital core using a four-parameter structure generation method. However, this method exhibits strong randomness in the generation of nanopores, making it difficult to accurately characterize the micro- and nanoscale pore structure information of shale.

[0057] In view of the above, in an exemplary embodiment, such as Figure 1 As shown, this application provides a method for digital core reconstruction at the nanometer-precision millimeter scale, including the following steps 101 to 106.

[0058] Step 101: Obtain the physical properties of the actual rock core.

[0059] Step 102: Obtain digital cores at different resolutions; different resolutions include micrometer resolution, submicrometer resolution and nanometer resolution; micrometer resolution digital cores are at the millimeter scale.

[0060] Step 103: Use a denoising diffusion probability model to augment digital core samples at different resolutions to obtain digital core datasets augmented for each resolution.

[0061] Step 104: Train a pSp super-resolution network using the digital core datasets augmented with two adjacent resolutions to obtain super-resolution reconstruction models with different resolutions.

[0062] Step 105: Using super-resolution reconstruction models with different resolutions, perform super-resolution reconstruction of the real micron-resolution digital core step by step to obtain the reconstructed nanometer-resolution millimeter-scale digital core.

[0063] Step 106: Based on the physical properties of the real core, correct the reconstructed nanometer-resolution millimeter-scale digital core to obtain the final nanometer-precision millimeter-scale digital core.

[0064] To implement steps 101 to 106 above, this application provides a joint super-resolution network framework based on DDPM (Denoising Diffusion Probabilistic Model)-pSp. By progressively increasing the resolution of digital cores, a cross-scale digital core with millimeter-scale and nanometer-scale resolution is constructed. The constructed digital core can not only identify the topological features of nanoscale pore structures but also reflect the structural information of micrometer-scale fractures.

[0065] In another exemplary embodiment of this application, a real rock core is obtained from the block, and the core is washed with oil and dried. The porosity, permeability, and pore size distribution curve of the core are measured experimentally. The process of obtaining the physical property parameters in step 101 can then be replaced by steps 201 to 204.

[0066] Step 201: Determine the physical property parameters, including porosity, permeability, and pore size distribution curve.

[0067] Step 202: Measure the core porosity using the helium expansion method.

[0068] First, helium gas is injected into the reference chamber to the initial pressure. Then the connecting valve is opened, and helium gas in the reference chamber expands into the sample chamber until the system reaches equilibrium pressure. According to Boyle's Law The volume of the sample skeleton can be calculated. :

[0069] ;

[0070] In the formula, For reference chamber volume, This represents the total volume of the sample chamber. Sample porosity. for: ,in This represents the external volume of the sample.

[0071] Step 203: Measure the permeability of the real core using the pulse attenuation method.

[0072] The pulse decay method is used to start timing from the moment the pressure pulse is applied, recording the pressure change over time to obtain the core pressure recovery curve. The horizontal axis of the core pressure recovery curve represents time, and the vertical axis represents pressure. The pressure recovery curve can be approximated as an exponential decay curve. By nonlinear fitting parameters Calculate the penetration rate .in, For a moment Pressure gradient over time, The gradient for pressure to return to the final equilibrium pressure. For penetration rate, It is fluid viscosity. It is the overall compressibility coefficient of the rock core. It is the core length.

[0073] Step 204: Measure the core pore size distribution curve using nuclear magnetic resonance.

[0074] The relaxation time (T2 spectrum) of fluid within pores in a magnetic field was measured using nuclear magnetic resonance (NMR); the relationship between the T2 spectrum and pore size was then analyzed. The pore size distribution curve of the sample was obtained, where For aperture, is the conversion factor.

[0075] In another exemplary embodiment of this application, in order to obtain scanned digital cores of different resolutions from real cores, the above step 102 can be replaced by the following steps 301 to 303.

[0076] Step 301: Scan the real rock core using a micron-scale CT device to obtain a digital rock core with micron-scale resolution and millimeter scale. Denoise the digital rock core with micron-scale resolution and millimeter scale to obtain a denoised digital rock core with micron-scale resolution and millimeter scale.

[0077] The core was cut into millimeter-sized rock samples and scanned using a micron-scale CT device to obtain a digital core with a resolution of N1 (5μm-20μm). The core was then denoised to obtain a denoised micron-scale resolution digital core DR1. The pore structure in DR1 is a micron-scale pore structure.

[0078] Step 302: Scan the real rock core using a nano-CT device to obtain a digital rock core with submicron resolution and micron scale. Denoise the digital rock core with submicron resolution and micron scale to obtain a denoised digital rock core with submicron resolution and micron scale.

[0079] A core sample measuring 200μm×200μm×200μm was cut from a millimeter-sized core sample. The core was then scanned using a nano-CT device to obtain a digital core with a resolution of N2 (50nm-400nm). After noise reduction processing, a submicron resolution digital core DR2 was obtained. The pore structure in DR2 is a submicron-level pore structure.

[0080] Step 303: Scan the real rock core using a focused ion beam scanning electron microscope to obtain a digital rock core with nanometer resolution and micrometer scale. Then, perform noise reduction and alignment on the digital rock core with nanometer resolution and micrometer scale to obtain the processed digital rock core with nanometer resolution and micrometer scale.

[0081] A small sample of 30μm×30μm×30μm was cut from the core sample. The core was scanned using a FIB-SEM device to obtain a digital core with a resolution of N3 (10nm-40nm). The core was then denoised and aligned to obtain the processed nanoscale resolution digital core DR3. The pore structure in DR3 is a nanoscale pore structure.

[0082] In another exemplary embodiment of this application, since a large amount of sample data is required when training the super-resolution network, the acquired first... Digital cores of various resolutions To expand. enter ( The corresponding denoising diffusion probability model is used for training to obtain... of The model; then using the trained... The model is augmented with a digital core dataset to generate training data for the super-resolution network. Therefore, step 103 above can be replaced by steps 401 to 402.

[0083] Step 401: Train a denoising diffusion probability model separately using digital cores of each resolution to obtain the trained denoising diffusion probability models corresponding to different resolutions.

[0084] The denoising diffusion probability model uses U-Net as its core network architecture, which includes a forward denoising module and a backward denoising module.

[0085] The forward noise-adding module progressively adds Gaussian noise to the digital core until the core becomes completely noisy. The core with added noise is represented by the following formula:

[0086] ;

[0087] In the formula, To add noise The digital core samples, ∈{1, 2, 3, …, T}, This indicates the total number of noise additions. To add noise The noise attenuation coefficient of the second order. , As The parameter increases with the increase of; The input is a digital core sample; This indicates that the mean is 0 and the variance is 0. Gaussian distribution; This is added noise.

[0088] The inverse denoising module uses a trained U-Net to predict the noise contained in the core sample for each denoising iteration. The input is then subtracted to gradually restore the true digital core. U-Net employs a symmetrical encoder-decoder structure. The encoder consists of four residual blocks, each including a convolutional layer, normalization, and activation function. It extracts high-dimensional features by progressively reducing the resolution of the input digital core. The decoder uses transposed convolutions to progressively restore the resolution of the digital core. The input to U-Net is a noisy digital core. and the number of times noise is added The output is a prediction noise with the same shape as the input core. .

[0089] Number of noise additions The encoding is used as an embedding vector, which is input to each residual block of the encoder. The output features of each residual block of the encoder are directly output to the corresponding layer of the decoder through skip connections to compensate for the information loss caused by downsampling.

[0090] Using sinusoidal position coding to determine the number of noise additions Encode as an input vector:

[0091] ;

[0092] In the formula, For each dimension of the encoding vector, This represents the total dimension of the encoded vector. The input vector is used. A multilayer perceptron (MLP) is used to map the input vector to obtain the embedding vector.

[0093] DDPM is trained using the following steps:

[0094] First, a core sample is randomly selected from the scanned core. And randomly select the number of times noise is added. ;

[0095] A forward noise-adding module is used to add noise to the core, resulting in a noisy digital core. ;

[0096] Will and number of noise additions Input U-Net, U-Net predicts the noise added in the previous step. ;

[0097] Calculate prediction noise and added noise The mean square error between (real noise) and the U-Net is updated through backpropagation. The formula for calculating the mean square error is:

[0098] ;

[0099] In the formula, For loss function, For the weights and biases of the U-Net network, This represents a standard Gaussian distribution with a mean of 0 and a variance of 1. Mean square error, Let be the mathematical expectation.

[0100] If the loss function is higher than the set value, repeat the above steps to continue training; when the loss function is lower than the set value, stop training and obtain the trained denoising diffusion probability model. .

[0101] Step 402: Based on the digital cores of each resolution, generate multiple digital cores using the trained denoising diffusion probability model corresponding to each resolution, and combine the multiple digital cores to form a digital core dataset expanded for each resolution.

[0102] Use the training The model is augmented with digital core data. First, pure noise of the same size as the target digital core is initialized, then... =500 to =1 iteration noise reduction, the specific form is as follows:

[0103] ;

[0104] In the formula, To add noise Digital core samples; It is random noise. ; The standard deviation of random noise, , For a moment The noise attenuation coefficient. When When =1, output the generated digital core. Repeat this process to generate multiple digital cores.

[0105] In another exemplary embodiment of this application, a low-resolution digital core dataset augmented with two adjacent resolutions is used. and high-resolution digital core dataset Train a pSp super-resolution network. The pSp super-resolution network consists of two structures: an encoder and a StyleGAN generator. Low-resolution digital core samples are cropped and used as input to the encoder; the cropped core sample size is equal to the original core sample size. , This represents a super-resolution multiplier; low-resolution digital core samples are processed by an encoder and a StyleGAN generator to output high-resolution digital core samples. The specific expression is as follows:

[0106] ;

[0107] In the formula, This is a cropped, low-resolution digital core sample. For StyleGAN generator; For encoder; is the average latent vector of the pre-trained generator.

[0108] The encoder contains a three-level feature extraction module for extracting multi-scale feature maps (including 3 large-scale feature maps, 4 medium-scale feature maps and 11 small-scale feature maps) from the input digital core. Each feature extraction module consists of three convolutional layers.

[0109] The extracted feature maps are progressively downsampled using a map2style intermediate mapping network (three convolutional layers with a stride of 2 and a LeakyReLU activation function) until the feature maps are compressed into 18 512-dimensional style vectors. , which serves as the input to the StyleGAN generator.

[0110] The StyleGAN generator consists of 18 convolutional blocks, with an initial input tensor and 18 style vectors as inputs. The initial input tensor is a constant 4×4×4 tensor, and the 18 style vectors are used as inputs to the 18 convolutional blocks respectively. The output of StyleGAN is a high-resolution digital core sample.

[0111] Each convolutional block consists of an upsampling layer, a convolutional layer, an AdanIN (Adaptive Instance Normalization) layer, and a LeakyReLU layer in sequence; the AdanIN layer receives the output feature map from the previous layer and modifies the input style vector. Perform affine transformation ( ), to embed feature style vectors, where For model parameters, For bias, This is the output style vector. Output style vectors for the encoder.

[0112] The training process for a super-resolution network is as follows:

[0113] First, from low-resolution digital core datasets Randomly select a rock core and input it into... In a super-resolution network, the output is a high-resolution digital core sample.

[0114] Next, the loss between the generated high-resolution digital core and the real high-resolution digital core is calculated:

[0115] ;

[0116] In the formula, for Loss during the training process of super-resolution networks To generate high-resolution digital cores, For true high-resolution digital cores; , , All are loss weighting coefficients; The pixel-level reconstruction loss is used to calculate the mean square error between the generated digital core and the real high-resolution digital core:

[0117] ;

[0118] In the formula, 、 and These are the length, height, and width of the input digital core. For generating high-resolution digital cores In position pixel values, For true high-resolution digital core In position The pixel value.

[0119] To detect similarity loss, a pre-trained VGG network is used to extract features from the generated high-resolution digital core and the real high-resolution digital core at different levels, and the feature similarity between the generated digital core and the real high-resolution digital core is calculated:

[0120] ;

[0121] In the formula, The first extracted for the VGG network Layer features; This is the normalization factor. The VGG network is a deep neural network composed of multiple repeatedly stacked small convolutional kernels. Pre-trained VGG networks are often used to extract high-level features of images to calculate perceptual loss.

[0122] The regularization loss is used to constrain the encoder's output style vector. Approximate StyleGAN average vector Improve the quality and output stability of generated digital cores:

[0123] ;

[0124] In the formula, The style vector output by the encoder.

[0125] Therefore, step 104 above can specifically include: forming a resolution pair between micron and submicron resolution digital cores, and forming a resolution pair between submicron and nanometer resolution digital cores; using the expanded micron resolution digital cores and the expanded submicron resolution digital cores as training sets to train the pSp super-resolution network, thereby obtaining a digital core super-resolution reconstruction model from micron to submicron resolution; using the expanded submicron resolution digital cores and the expanded nanometer resolution digital cores as training sets to train the pSp super-resolution network, thereby obtaining a digital core super-resolution reconstruction model from submicron to nanometer resolution.

[0126] In another exemplary embodiment of this application, cross-scale digital core construction: the digital core DR1 with a resolution of N1μm is divided into... Smaller sizes of the same dimensions (size DR1) A digital core sample with a resolution of N = 1 μm was obtained. A trained super-resolution reconstruction model was used for super-resolution reconstruction, resulting in a digital core sample DM2 with a resolution of N = 2 nm. The digital core sample DM2 was then divided into... Smaller sizes of the same dimensions (size DM2) The core sample is reconstructed into a digital core DM3 with a resolution of N3nm using a trained super-resolution reconstruction model. Step 105 can then be replaced by steps 501-507.

[0127] Step 501: Determine the super-resolution reconstruction models with different resolutions, including submicron resolution super-resolution reconstruction models and nanometer resolution super-resolution reconstruction models.

[0128] Step 502: Divide the digital core at the true micron resolution and millimeter scale into... There are 10 core samples of the same size. n is a positive integer.

[0129] Step 503: Based on the core samples divided in Step 502, obtain the sub-micron resolution super-resolution reconstruction model. A reconstructed digital core with submicron resolution.

[0130] Step 504: Combine all the reconstructed submicron resolution digital cores according to the division order in Step 502 to construct a submicron resolution millimeter-scale digital core.

[0131] Step 505: Divide the reconstructed submicron resolution millimeter-scale digital core into... Three core samples of the same size. m is a positive integer.

[0132] Step 506: Based on the core samples divided in Step 505, obtain the super-resolution reconstruction model using nanometer resolution. A reconstructed digital core at nanometer resolution.

[0133] Step 507: Combine all the reconstructed nanometer-resolution digital cores according to the division order in Step 505 to construct a nanometer-resolution millimeter-scale digital core.

[0134] In another exemplary embodiment of this application, the digital core model is corrected as follows: Based on the experimentally measured porosity, an appropriate threshold is selected to divide the core skeleton and pores; the core porosity is calculated and compared with the experimental measurement results, with an error of less than 1%. The permeability and pore size distribution are simulated and calculated using a pore network, and compared with the experimental results. If the calculation error is less than 5%, the established digital core meets the requirements; otherwise, the pixels at the edges of the pores and skeleton are modified, and the permeability and pore size distribution are re-simulated and calculated until the calculation error is less than 5%. Step 106 can then be replaced by steps 601 to 606.

[0135] Step 601: Determine the porosity threshold based on the porosity of the actual core.

[0136] Step 602: Divide the skeleton and pores of the reconstructed nanometer-resolution millimeter-scale digital core according to the pore threshold, and construct a three-dimensional pore network model of the core.

[0137] Step 603: Calculate the pore size distribution based on the pore network model of the core; and classify the pores into small pores and large pores according to the pore size.

[0138] Step 604: If the error between the simulated pore size distribution and the actual pore size distribution of the core is less than a preset threshold, then calculate the permeability of the three-dimensional pore network model of the core.

[0139] Step 605: If the error between the calculated permeability and the permeability of the actual core is less than a preset threshold, then the reconstructed nanometer-resolution millimeter-scale digital core is determined as the final nanometer-precision millimeter-scale digital core.

[0140] Step 606: If the error between the simulated pore size distribution and the actual pore size distribution of the core is greater than or equal to a preset threshold, then modify the pixels at the edges of the pores and skeleton in the reconstructed nanometer-resolution millimeter-scale digital core, and return to step 602.

[0141] If the calculated macropore frequency distribution is higher than the experimental results and the calculated permeability is higher than the actual value, then image erosion is performed on the macropores; otherwise, image dilation is performed on the macropores. If the calculated micropore frequency distribution is higher than the experimental results and the calculated permeability is higher than the actual value, then image erosion is performed on the micropores; otherwise, image dilation is performed on the micropores.

[0142] The method described in this application will be illustrated below using an actual shale oil reservoir core as an example.

[0143] S1: Obtaining Authentic Core Physical Properties: Obtain authentic core samples from the block, wash and dry the core samples, and experimentally measure the porosity of the core samples. Penetration rate and aperture distribution curve .

[0144] S2: Acquiring multi-resolution imaging data of rock cores: Rock cores are scanned using micro / nano CT equipment and FIB-SEM equipment to obtain digital rock cores at different resolutions, such as... Figure 2 As shown. Figure 2 The 5mm, 0.2mm, and 10μm values ​​are all scale bars.

[0145] S21: Core scans were performed using a micron-scale CT scanner to obtain images with a resolution of 10 μm and a voxel count of 800. 3 The digital core DR1 was obtained and its noise was reduced to obtain... Figure 2 The digital core with micron resolution shown in section (a) includes a micron-scale pore structure. DR1 measures 8 mm × 8 mm × 8 mm.

[0146] S22: The core was scanned using a nano-CT device to obtain a digital core DR2 with a resolution of 400nm, and noise reduction processing was performed to obtain... Figure 2 The submicron resolution digital core shown in section (b) includes a submicron-scale pore structure.

[0147] S23: The core was scanned using a FIB-SEM device to obtain a digital core DR3 with a resolution of 20nm. Noise reduction and alignment were then performed to obtain... Figure 2 The digital core shown in section (c) has a nanoscale resolution and includes a nanoscale pore structure.

[0148] S3: Core Sample Enlargement: Construction Based on U-Net Network Model, core enter Training was conducted to obtain rock cores. of The model; then the trained DDPM is used. i Model for core Perform sample augmentation to generate training data for the super-resolution network. . Figure 3 On the left are the DDPM forward noise addition module and the reverse noise reduction module. The number of noise additions is The digital core image is shown on the right; the U-Net network structure is shown on the right.

[0149] 10,000 of size 800 3 The noise vector is input into the trained DDPM1 to generate 10,000 digital core DRS1 images with a resolution of 10 μm; 10,000 images of size 800 are then processed. 3 The noise vector is input into the trained DDPM2 to generate 10,000 digital core DRS2 images with a resolution of 400 nm. The 10,000 images are then processed into 800-bit digital core images. 3 The noise vector is input into the trained DDPM3 to generate 10,000 digital core DRS3s with a resolution of 20nm. Figure 4 A schematic diagram of digital cores generated for DDPM is shown. The left side of the arrow represents the training data, and the right side represents the generated digital core.

[0150] S4: Training the super-resolution network. Figure 5 This is a schematic diagram of the pSp super-resolution network. The structure within the dashed box on the left is the encoder, the structure within the dashed box on the right is the StyleGAN generator, and map2style is the intermediate mapping network.

[0151] DRS1 and DRS2 are input into the super-resolution network pSp2 to train the super-resolution reconstruction model SR2; DRS2 and DRS3 are input into the super-resolution network pSp3 to train the super-resolution reconstruction model SR. 34 . Figure 6 This is a schematic diagram of the pSp network training results. Figure 6 Part (a) shows a schematic diagram of the training results of the super-resolution network pSp2. Figure 6 Part (b) shows a schematic diagram of the training results of the super-resolution network pSp3.

[0152] S5: Cross-scale digital core construction: Using the obtained super-resolution reconstruction models SR2 and SR3, low-resolution digital cores are gradually constructed. Reconstructed into high-resolution digital cores, such as Figure 7 As shown.

[0153] S51: Divide the digital core DR1 into 10 3 Two identical samples (each sample size is 80). 3 (Voxels), and input them sequentially into SR2 to obtain 10 3 Individual statistic 2000 3 S52: Combine all the reconstructed submicron resolution digital cores DM2 according to the subsample division order in step S51 to construct a submicron resolution millimeter-scale digital core.

[0154] S53: Divide the reconstructed submicron resolution millimeter-scale digital core into 10... 6 Two identical samples (each sample size is 200). 3 (Voxels), and input them sequentially into SR3 to obtain 10 6 Individual statistic is 4000 3 Digital core DM3 with a resolution of 20nm nanometer resolution.

[0155] S54: Combine all the reconstructed nanoscale resolution digital cores DM3 according to the subsample division order in step S53 to construct a nanoscale resolution millimeter-scale digital core.

[0156] S6: Digital Core Model Correction: Based on experimentally measured porosity, select an appropriate threshold to delineate the framework and pores of core DM3; use the pore network to simulate and calculate its permeability and pore size distribution, and compare the results with experimental results. If the calculation error is less than 5%, the established digital core meets the requirements; otherwise, modify the pixels at the edges of the pore framework and recalculate the permeability and pore size distribution until the calculation error is less than 5%. Figure 8 As shown, the measured porosity is 5.3% and the permeability is 0.165 mD; the porosity obtained through simulation calculation is 5.25% and the permeability is 0.158 mD; the error of the pore size distribution curve is 3.66%, and the errors are all less than the set threshold.

[0157] S61: Using threshold segmentation, the reconstructed core porosity is calculated to ensure that the absolute error between the calculation result and the experimental measurement value is less than 1%.

[0158] S62: Based on pore size, pores are divided into micropores and macropores. The edge pixels of the corresponding pores are modified according to the difference in pore size distribution to ensure that the error between the calculated pore size distribution and permeability and the experimental measurement value is less than 5%. If the calculated frequency distribution of macropores is higher than the experimental result and the calculated permeability is higher than the actual value, then image erosion is performed on the macropores; otherwise, image dilation is performed on the macropores. If the calculated frequency distribution of micropores is higher than the experimental result and the calculated permeability is higher than the actual value, then image erosion is performed on the micropores; otherwise, image dilation is performed on the micropores.

[0159] This application first uses experimental methods to obtain the physical properties (porosity, permeability, pore size distribution) of real rock cores. Then, it uses micron-scale CT, nano-scale CT, and FIB-SEM to scan and obtain digital rock core data at different resolutions. The rock core samples are expanded using a denoising diffusion probability model to generate training samples for a super-resolution network. Using pairs of digital rock core data at adjacent resolutions, the pSp super-resolution network is trained to obtain super-resolution reconstruction models at different resolutions. Using these super-resolution reconstruction models, the resolution of the digital rock cores is progressively increased from micron to submicron to nanometer, constructing a nanometer-precision millimeter-scale digital rock core. Finally, the digital rock cores are corrected based on experimental data to ensure that the errors in key parameters such as porosity and permeability are controlled within 5%. Due to the limitations of current scanning equipment precision, existing experimental methods cannot simultaneously obtain digital rock cores at both nanometer and millimeter scales, making it difficult to accurately characterize the multi-scale pore and fracture structure of shale. The method in this application overcomes the problem of the contradiction between resolution and field of view in existing imaging experiments.

[0160] This application innovatively integrates deep learning and multi-resolution imaging technologies, resolving the contradiction between resolution and field of view in traditional methods. The constructed digital core exhibits nanometer-level precision in pore structure characterization and millimeter-level dimensions, enabling the observation of micrometer-scale fractures. This provides crucial technical support for accurately characterizing the multi-scale pore and fracture structure features and seepage mechanisms of unconventional reservoirs.

[0161] Based on the same inventive concept, this application also provides a nanometer-precision millimeter-scale digital core reconstruction device for implementing the aforementioned nanometer-precision millimeter-scale digital core reconstruction method. The solution provided by this device is similar to the implementation described in the above method. Therefore, the specific limitations of one or more nanometer-precision millimeter-scale digital core reconstruction device embodiments provided below can be found in the limitations of the nanometer-precision millimeter-scale digital core reconstruction method described above, and will not be repeated here.

[0162] In one exemplary embodiment, such as Figure 9As shown, a nanometer-precision millimeter-scale digital core reconstruction device is provided, comprising: a parameter acquisition module, a digital core acquisition module, a digital core sample expansion module, a training module, a reconstruction module, and a correction module.

[0163] The system comprises the following modules: a parameter acquisition module for acquiring the physical properties of real core samples; a digital core acquisition module for acquiring digital core samples at different resolutions, including micrometer, submicrometer, and nanometer resolutions; and a micrometer resolution digital core sample for millimeter-scale core samples. A digital core sample expansion module is used to expand the digital core samples at different resolutions using a denoising diffusion probability model, obtaining expanded digital core datasets for each resolution. A training module trains a pSp super-resolution network using the expanded digital core datasets at two adjacent resolutions, obtaining super-resolution reconstruction models at different resolutions. A reconstruction module uses these super-resolution reconstruction models to progressively reconstruct the real micrometer resolution digital core samples, obtaining reconstructed nanometer resolution millimeter-scale digital core samples. A correction module corrects the reconstructed nanometer resolution millimeter-scale digital core samples based on the physical properties of the real core samples, obtaining the final nanometer precision millimeter-scale digital core samples.

[0164] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 10 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores the final nanometer-precision millimeter-scale digital core data. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a nanometer-precision millimeter-scale digital core reconstruction method.

[0165] Those skilled in the art will understand that Figure 10The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0166] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0167] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0168] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0169] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0170] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0171] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0172] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for digital core reconstruction at the nanometer-precision millimeter scale, characterized in that, include: Obtain the physical properties of real rock cores; Digital core samples were acquired at different resolutions. Different resolutions include micrometer resolution, submicrometer resolution, and nanometer resolution; Digital core samples with micrometer resolution are at the millimeter scale. A denoising diffusion probability model was used to augment digital core samples at different resolutions, resulting in digital core datasets augmented for each resolution. A pSp super-resolution network was trained using digital core datasets augmented with two adjacent resolutions to obtain super-resolution reconstruction models at different resolutions. By using super-resolution reconstruction models with different resolutions, real micron-resolution digital cores are reconstructed step by step to obtain reconstructed nanometer-resolution millimeter-scale digital cores. Based on the physical properties of the real core, the reconstructed nanometer-resolution millimeter-scale digital core is corrected to obtain the final nanometer-precision millimeter-scale digital core. A pSp super-resolution network was trained using digital core datasets augmented with two adjacent resolutions to obtain super-resolution reconstruction models at different resolutions, specifically including: Define any two adjacent resolutions in different resolutions as a resolution pair; Using the augmented digital cores corresponding to the low and medium resolutions of each resolution pair as input and the augmented digital cores corresponding to the medium and high resolutions of each resolution pair as output, a pSp super-resolution network is trained to obtain the super-resolution reconstruction model for each resolution pair. The super-resolution reconstruction models for all resolution pairs are then combined to form super-resolution reconstruction models for different resolutions. Using super-resolution reconstruction models of different resolutions, real micron-resolution digital cores are progressively reconstructed to obtain reconstructed nanometer-resolution millimeter-scale digital cores, specifically including: Super-resolution reconstruction models with different resolutions were determined, including submicron resolution super-resolution reconstruction models and nanometer resolution super-resolution reconstruction models; Divide the digital core sample with true micron resolution and millimeter scale into n 3 A set of core samples of the same size; n is a positive integer; Based on the core samples divided at micrometer resolution, a super-resolution reconstruction model at submicrometer resolution is used to obtain n. 3 A reconstructed digital core with submicron resolution; All the reconstructed submicron resolution digital cores were combined according to the division order of core samples at micron resolution to construct a submicron resolution millimeter-scale digital core. The reconstructed submicron resolution millimeter-scale digital core was divided into m... 3 A set of core samples of the same size; m is a positive integer; Based on the core samples divided at submicron resolution, a super-resolution reconstruction model at nanometer resolution was used to obtain m 3 A reconstructed digital core at nanometer resolution; All the reconstructed nanometer-resolution digital cores were combined according to the sub-micrometer resolution core sample division order to construct a nanometer-resolution millimeter-scale digital core.

2. The nanometer-precision millimeter-scale digital core reconstruction method according to claim 1, characterized in that, Obtaining the physical properties of real rock cores specifically includes: The physical properties include porosity, permeability, and pore size distribution curves; Core porosity was measured using the helium expansion method. The permeability of real rock cores was measured using the pulse attenuation method. The core pore size distribution curve was measured using nuclear magnetic resonance.

3. The nanometer-precision millimeter-scale digital core reconstruction method according to claim 1, characterized in that, Digital core samples were acquired at different resolutions, specifically including: Real rock cores were scanned using a micron-CT device to obtain digital rock cores with micron resolution and millimeter scale. The digital rock cores with micron resolution and millimeter scale were then denoised to obtain denoised digital rock cores with micron resolution and millimeter scale. A real rock core was scanned using a nano-CT device to obtain a digital rock core with submicron resolution and micron scale. The submicron resolution and micron scale digital rock core was then denoised to obtain a denoised submicron resolution and micron scale digital rock core. A digital core at nanometer resolution and micrometer scale was obtained by scanning a real rock core using a focused ion beam scanning electron microscope. The digital core at nanometer resolution and micrometer scale was then denoised and aligned to obtain a processed digital core at nanometer resolution and micrometer scale.

4. The nanometer-precision millimeter-scale digital core reconstruction method according to claim 1, characterized in that, A denoising diffusion probability model was used to augment digital core samples at different resolutions, resulting in digital core datasets for each resolution. Specifically, these datasets include: A separate denoising diffusion probability model was trained using digital cores of each resolution to obtain the trained denoising diffusion probability models corresponding to different resolutions. Based on the digital cores at each resolution, multiple digital cores are generated using the pre-trained denoising diffusion probability model corresponding to each resolution, and these multiple digital cores are combined to form a digital core dataset expanded for each resolution.

5. The nanometer-precision millimeter-scale digital core reconstruction method according to claim 2, characterized in that, Based on the physical properties of the actual core, the reconstructed nanometer-resolution millimeter-scale digital core is corrected to obtain the final nanometer-precision millimeter-scale digital core, specifically including: Determine the porosity threshold based on the porosity of the actual rock core; The reconstructed nanometer-resolution millimeter-scale digital core is divided into its skeleton and pores according to the pore threshold, and a three-dimensional pore network model of the core is constructed. Based on the three-dimensional pore network model of the rock core, the pore size distribution is calculated; If the error between the simulated pore size distribution and the actual pore size distribution of the core is less than a preset threshold, then the permeability of the three-dimensional pore network model of the core is calculated. If the error between the calculated permeability and the permeability of the actual core is less than a preset threshold, the reconstructed nanometer-resolution millimeter-scale digital core will be determined as the final nanometer-precision millimeter-scale digital core. If the error between the simulated pore size distribution and the actual pore size distribution of the core is greater than or equal to a preset threshold, then the pixels at the edges of the pores and skeleton in the reconstructed nanometer-resolution millimeter-scale digital core are modified, and the process returns to the step of dividing the skeleton and pores of the reconstructed nanometer-resolution millimeter-scale digital core according to the pore threshold to construct a three-dimensional pore network model of the core.

6. A nanometer-precision millimeter-scale digital core reconstruction device, characterized in that, The nanometer-precision millimeter-scale digital core reconstruction device is used to implement the nanometer-precision millimeter-scale digital core reconstruction method according to any one of claims 1-5, and the nanometer-precision millimeter-scale digital core reconstruction device comprises: The parameter acquisition module is used to acquire the physical property parameters of real rock cores; The digital core acquisition module is used to acquire digital cores at different resolutions, including micrometer, submicrometer, and nanometer resolutions; the micrometer resolution digital core is at the millimeter scale. The digital core sample augmentation module is used to augment digital core samples at different resolutions using a denoising diffusion probability model, thereby obtaining digital core datasets augmented at each resolution. The training module is used to train a pSp super-resolution network using digital core datasets augmented with two adjacent resolutions, to obtain super-resolution reconstruction models with different resolutions. The reconstruction module is used to perform super-resolution reconstruction of real micron-resolution digital cores step by step using super-resolution reconstruction models of different resolutions, so as to obtain reconstructed nanometer-resolution millimeter-scale digital cores. The correction module is used to correct the reconstructed nanometer-resolution millimeter-scale digital core based on the physical property parameters of the real core, so as to obtain the final nanometer-precision millimeter-scale digital core.

7. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the nanometer-precision millimeter-scale digital core reconstruction method according to any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the nanometer-precision millimeter-scale digital core reconstruction method as described in any one of claims 1-5.

Images