CN-120655814-B - Weathered sandstone microscopic image generation method based on condition generation countermeasure network

CN120655814BCN 120655814 BCN120655814 BCN 120655814BCN-120655814-B

Abstract

The invention relates to the technical field of digital core modeling and artificial intelligence modeling, in particular to a method for generating a weathered sandstone microscopic image based on a condition generation countermeasure network, which comprises the steps of acquiring and analyzing sandstone samples with different weathered degrees to construct a weathered sandstone multi-scale database; the method comprises the steps of carrying out weathered geological knowledge extraction on a weathered sandstone multi-scale database to construct a sandstone characteristic tag set, training a condition generation countermeasure network by utilizing the sandstone characteristic tag set to obtain a weathered sandstone microscopic image generation model, and further generating a weathered sandstone microscopic structure image with physical consistency by utilizing the weathered sandstone microscopic image generation model. Therefore, the problems that the depth association relation between the microstructure and the macroscopic mechanical property is difficult to reveal in the conventional weathered sandstone physical mechanical property modeling, the advantages of the weathered geological mechanism and the generated modeling are difficult to be effectively fused by the conventional modeling method, the physical consistency and the weathered response rationality are lacked in the reconstruction result are solved.

Inventors

Qiao Jiangmei
TANG XUHAI
XU SHENGZONG
TONG YUWEN

Assignees

武汉大学

Dates

Publication Date: 20260505
Application Date: 20250421

Claims (9)

1. A method for generating a weathered sandstone microscopic image based on a conditional generation antagonism network, comprising the steps of: Acquiring and analyzing sandstone samples with different weathering degrees to construct a weathered sandstone multi-scale database; Extracting weathered geological knowledge from the weathered sandstone multi-scale database to construct a sandstone characteristic tag set, which specifically comprises the following steps: extracting mineral composition data, weathered product data and microstructure data in the weathered sandstone multi-scale database; Quantitatively analyzing the mineral composition data to obtain corresponding volume fraction and mass fraction, and setting mineral composition labels of the sandstone samples with different weathering degrees according to the volume fraction and the mass fraction; Setting the weathering degree labels of the sandstone samples with different weathering degrees according to the weathering product data based on a preset weathering grade standard; Extracting microstructure features in the microstructure data, and setting microstructure feature labels of sandstone samples with different weathering degrees according to the microstructure features; Analyzing the mineral crystal orientation distribution in the microstructure data, and setting the crystal orientation degree labels of the sandstone samples with different weathering degrees according to the mineral crystal orientation distribution; Calculating the similarity of local details of sandstone mineral crystals in the microstructure data to obtain corresponding structural similarity, and setting structural similarity labels of sandstone samples with different weathering degrees according to the structural similarity; Preprocessing the mineral component label, the weathering degree label, the microstructure feature label, the crystal orientation degree label and the structural similarity label to construct a sandstone feature label set with real physical properties; Training a pre-constructed condition generation countermeasure network by utilizing the sandstone characteristic tag set so as to obtain a weathered sandstone microscopic image generation model; Inputting the sandstone characteristic label to be predicted into the weathered sandstone microscopic image generating model to generate a weathered sandstone microscopic structure image.
2. The method of generating a microimage of weathered sandstone based on a condition generating antagonism network of claim 1, wherein said acquiring and analyzing sandstone samples of different degrees of weathering to construct a weathered sandstone multi-scale database comprises: collecting sandstone samples with different weathering degrees; Acquiring microscopic parameters of the sandstone samples with different weathering degrees by using a preset diffraction device; obtaining macroscopic physical and mechanical parameters of the sandstone samples with different weathering degrees through macroscopic physical experiments; And carrying out data processing on the microscopic parameters and the macroscopic physical and mechanical parameters to obtain a weathered sandstone multi-scale database with the corresponding relation between the weathered stage labels and the structural parameters.
3. The method for generating a weathered sandstone microscopic image based on a condition generating countermeasure network according to claim 1, wherein the training of the pre-constructed condition generating countermeasure network by using the sandstone feature tag set to obtain a weathered sandstone microscopic image generating model comprises: Inputting the sandstone characteristic tag set and the sandstone sample image corresponding to the sandstone characteristic tag set into a pre-built condition generation countermeasure network based on an Adam optimizer, and carrying out iterative updating on parameters in the condition generation countermeasure network until a preset comprehensive loss function converges to obtain the weathered sandstone microscopic image generation model, wherein the preset comprehensive loss function comprises a countermeasure loss function, a mineral composition loss function, a direction loss function, a weathered product and pore ratio loss function and a structural similarity loss function.
4. The method of claim 1, wherein the pre-built condition generation countermeasure network comprises a generator and a discriminator, wherein the generator generates weathered sandstone microstructure images under corresponding weathered phases with random noise and the sandstone feature tag set as condition inputs, and the discriminator adopts PatchGAN architecture to distinguish weathered sandstone microstructure sample images from weathered sandstone microstructure images.
5. The method of generating a weathered sandstone microimage based on a condition generating challenge network of claim 1, further comprising: and comparing the weathered sandstone microstructure image with the weathered sandstone microstructure sample image corresponding to the sandstone characteristic label to be predicted, so as to quantitatively evaluate the performance of the weathered sandstone microstructure image generation model.
6. A weathered sandstone microscopic image generation device based on a condition generation antagonism network, comprising: the database construction module is used for acquiring and analyzing sandstone samples with different weathering degrees so as to construct a weathered sandstone multi-scale database; The tag set construction module is used for extracting weathered geological knowledge from the weathered sandstone multi-scale database to construct a sandstone characteristic tag set, and specifically comprises the following steps: the extraction unit is used for extracting mineral composition data, weathered product data and microstructure data in the weathered sandstone multi-scale database; The first setting unit is used for quantitatively analyzing the mineral composition data to obtain corresponding volume fraction and mass fraction, and setting mineral composition labels of sandstone samples with different weathering degrees according to the volume fraction and the mass fraction; The second setting unit is used for setting the weathering degree labels of the sandstone samples with different weathering degrees according to the weathering product data based on the preset weathering grade standard; The third setting unit is used for extracting microstructure features in the microstructure data and setting microstructure feature labels of sandstone samples with different weathering degrees according to the microstructure features; The fourth setting unit is used for analyzing the mineral crystal orientation distribution in the microstructure data and setting the crystal orientation degree labels of the sandstone samples with different weathering degrees according to the mineral crystal orientation distribution; the fifth setting unit is used for calculating the similarity of the local details of the sandstone mineral crystals in the microstructure data so as to obtain corresponding structural similarity, and setting structural similarity labels of sandstone samples with different weathering degrees according to the structural similarity; The construction unit is used for preprocessing the mineral component label, the weathering degree label, the microstructure characteristic label, the crystal orientation label and the structural similarity label to construct a sandstone characteristic label set with real physical properties; The training module is used for training a pre-constructed condition generation countermeasure network by utilizing the sandstone characteristic tag set so as to obtain a weathered sandstone microscopic image generation model; The generation module is used for inputting the sandstone characteristic label to be predicted into the weathered sandstone microscopic image generation model so as to generate a weathered sandstone microscopic structure image.
7. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor executing the program to implement the conditional generation opposing network based weathered sandstone microimage generation method of any of claims 1-5.
8. A computer program product, characterized in that the computer program/instructions, when executed by a processor, implement the conditional generation opposing network based weathered sandstone microscopic image generation method of any of claims 1 to 5.
9. A computer readable storage medium having stored thereon a computer program, wherein the program is executed by a processor for implementing the conditional generation opposing network based weathered sandstone microimage generation method of any of claims 1-5.

Description

Weathered sandstone microscopic image generation method based on condition generation countermeasure network Technical Field The invention relates to the technical field of digital core modeling and artificial intelligence modeling, in particular to a weathered sandstone microscopic image generation method based on a condition generation countermeasure network. Background Sandstones are widely distributed in natural environments, and the weathering behavior of sandstones has important effects on geological engineering, geological disaster prediction and rock stability research. In the weathering process, the mineral composition and microstructure (such as pores and cracks) of sandstone are obviously changed along with the time evolution, and the physical and mechanical properties of the sandstone are affected. The traditional weathering research method such as experimental observation and numerical simulation has the problems of long data acquisition period, high cost, weak response capability to complex environments and the like, and is difficult to reflect the continuous evolution process of mineral components and microstructures. In recent years, generation of an antagonistic neural Network (GAN) is excellent in the image field, and can make up for the shortage of limited experimental samples. However, the generic GAN model has a certain limitation in the control of complex conditions (such as rock weathering degree), and image generation based on the control conditions such as the weathering degree cannot be achieved yet. Therefore, development of a GAN model with combination of weathering degree is needed to solve the problems of prediction and reconstruction of microscopic image evolution of weathered rock. Disclosure of Invention The invention provides a method for generating a weathered sandstone microscopic image based on a conditional generation countermeasure network, which aims to solve the problems that a high-precision, multi-scale and physically significant weathered sandstone database is lacked in the existing weathered sandstone physical and mechanical property modeling, the deep association relationship between a microstructure and macroscopic mechanical properties is difficult to reveal, the traditional modeling method is difficult to effectively fuse the advantages of a weathered geological mechanism and the generated modeling, the physical consistency and the weathered response rationality are lacked in a reconstruction result, and the like. The embodiment of the first aspect of the invention provides a method for generating a weathered sandstone microscopic image based on a condition generation countermeasure network, which comprises the following steps of acquiring and analyzing sandstone samples with different weathered degrees to construct a weathered sandstone multi-scale database, extracting weathered geological knowledge of the weathered sandstone multi-scale database to construct a sandstone feature tag set, training the pre-constructed condition generation countermeasure network by utilizing the sandstone feature tag set to obtain a weathered sandstone microscopic image generation model, and inputting a sandstone feature tag to be predicted into the weathered sandstone microscopic image generation model to generate a weathered sandstone microscopic structure image. Optionally, the acquiring and analyzing sandstone samples with different weathering degrees to construct a weathered sandstone multi-scale database includes: the method comprises the steps of collecting sandstone samples with different degrees of weathering, obtaining microscopic parameters of the sandstone samples with different degrees of weathering by using preset diffraction equipment, obtaining macroscopic physical and mechanical parameters of the sandstone samples with different degrees of weathering by macroscopic physical experiments, and carrying out data processing on the microscopic parameters and the macroscopic physical and mechanical parameters to obtain a weathered sandstone multi-scale database with corresponding relations between labels of weathering stages and structural parameters. Optionally, the extracting weathered geological knowledge from the weathered sandstone multi-scale database to construct a sandstone feature tag set includes: The method comprises the steps of extracting mineral composition data, weathered product data and microstructure data in a weathered sandstone multi-scale database, quantitatively analyzing the mineral composition data to obtain corresponding volume fractions and quality fractions, setting mineral component labels of sandstone samples with different weathered degrees according to the volume fractions and the quality fractions, setting the weathered degree labels of the sandstone samples with different weathered degrees according to the weathered product data based on preset weathered grade standards, extracting microstructure features in the microstructure data, se