CN-120765869-B - Method, device, equipment and medium for constructing three-dimensional geologic model fusing multiple metadata

CN120765869BCN 120765869 BCN120765869 BCN 120765869BCN-120765869-B

Abstract

The invention discloses a three-dimensional geologic model construction method, device, equipment and medium for fusing multi-element data, which are applied to the field of geologic survey and comprise the steps of obtaining two-dimensional profile detection data of a target geologic region, expanding the two-dimensional profile detection data into three-dimensional training data, extracting space geologic constraint conditions, constructing a multi-element clustering model library according to the three-dimensional training data, a speed model and a density model, constructing an alternative model library according to a set simulation path and space geologic constraint conditions by utilizing the multi-element clustering model library, calculating according to the alternative model library to obtain a probability matrix, selecting modes according to the probability matrix and cross entropy to obtain a three-dimensional geologic initial model, and performing multi-scale iterative optimization on the three-dimensional geologic initial model to obtain an optimized three-dimensional geologic model. The method effectively improves the precision and reliability of the three-dimensional geological model, can better process the problems of data sparsity and geological complexity, and reduces the uncertainty of the simulation result.

Inventors

CHEN YONGHUA
WANG DIAN
LI YANHUA
LIAO JING
HOU WEISHENG
LIU JUN
HUANG YUHUI
YANG SONGHUA
GUO QINGFENG

Assignees

广州地铁设计研究院股份有限公司
中山大学

Dates

Publication Date: 20260505
Application Date: 20250616

Claims (8)

1. A three-dimensional geological model construction method integrating multiple data is characterized by comprising the following steps: Acquiring two-dimensional profile detection data of a target geological region, and expanding the two-dimensional profile detection data into three-dimensional training data; Extracting a space geological constraint condition from the two-dimensional profile detection data; constructing a multi-element clustering mode library according to the three-dimensional training data, the speed model and the density model; Constructing an alternative mode library by utilizing the multi-element clustering mode library according to the set simulation path and the space geological constraint condition; Calculating according to the alternative mode library to obtain a probability matrix, and selecting modes according to the probability matrix and cross entropy to obtain a three-dimensional geological initial model; Performing multi-scale iterative optimization on the three-dimensional geological initial model by adopting the three-dimensional training data to obtain an optimized three-dimensional geological model; And constructing an alternative mode library by utilizing the multi-element clustering mode library according to the set simulation path and the space geological constraint condition, wherein the method comprises the following steps of: Sequentially determining the positions of nodes to be simulated according to the set simulation paths; Extracting a plurality of modes associated with the positions of the nodes to be simulated from the multi-element clustering mode library for each node to be simulated; screening each mode according to the space geological constraint condition to obtain a target mode; Obtaining an alternative mode library according to the target modes of the nodes to be simulated; the method comprises the steps of calculating a probability matrix according to the alternative mode library, selecting modes according to the probability matrix and cross entropy, and obtaining a three-dimensional geological initial model, and comprises the following steps: Calculating the lithology data similarity of each target mode in the candidate mode library through a Hamming distance function, and calculating the speed data similarity and the density data similarity of each target mode in the candidate mode library through a Euclidean distance function; obtaining a similarity matrix according to the lithology data similarity, the speed data similarity and the density data similarity; normalizing the data in the similarity matrix to obtain a probability matrix; calculating probability weight coefficients of the target modes by using cross entropy; carrying out weight adjustment on the probability matrix according to each probability weight coefficient to obtain a comprehensive probability matrix; and selecting a target mode with the highest probability value from the comprehensive probability matrix, and collaging the target mode to the node to be simulated to obtain a three-dimensional geological initial model.
2. The method of constructing a three-dimensional geologic model incorporating multiple data of claim 1, wherein the two-dimensional profile detection data comprises geologic profile data, longitudinal wave velocity profile data, and density profile data; The expanding the two-dimensional profile detection data into three-dimensional training data comprises the following steps: and importing the geological profile data, the longitudinal wave speed profile data and the density profile data into a simulation grid, randomly expanding the simulation grid, and expanding the geological profile data, the longitudinal wave speed profile data and the density profile data into a three-dimensional profile with a certain width to obtain three-dimensional training data.
3. The method of constructing a three-dimensional geologic model that fuses multiple data according to claim 2, wherein the extracting spatial geologic constraints from the two-dimensional profile detection data comprises: extracting stratum sequence information and stratum thickness information from the geological profile data; identifying and obtaining the space contact relation between geological fault data and each geological body according to the geological profile data; And taking the stratum sequence information, stratum thickness information, geological fault data and spatial contact relation as the spatial geological constraint condition of the target geological region.
4. The method for constructing a three-dimensional geologic model that fuses multiple data according to claim 1, wherein the constructing a multi-dimensional cluster pattern library from the three-dimensional training data, the velocity model, and the density model comprises: constructing a multi-element mode library according to the three-dimensional training data, the speed model and the density model; and carrying out cluster analysis on the modes of the multi-mode library according to a similarity threshold value, and classifying the multi-mode library according to an analysis result to obtain the multi-cluster mode library.
5. The method for constructing a three-dimensional geologic model with multi-metadata fusion according to claim 1, wherein the performing multi-scale iterative optimization on the three-dimensional geologic initial model by using the three-dimensional training data to obtain an optimized three-dimensional geologic model comprises: Performing multi-scale division on the three-dimensional geological initial model to obtain a plurality of different resolution levels; performing iterative optimization on the three-dimensional geological initial model by adopting the three-dimensional training data on each resolution level, and upsampling the optimized three-dimensional geological initial model to the next resolution level; and repeating the iterative optimization process until the three-dimensional geological initial model reaches the highest resolution level, and obtaining the optimized three-dimensional geological model.
6. A three-dimensional geologic model construction device fusing multiple data, comprising: The data expansion module is used for acquiring two-dimensional profile detection data of the target geological region and expanding the two-dimensional profile detection data into three-dimensional training data; The constraint extraction module is used for extracting space geological constraint conditions from the two-dimensional profile detection data; the pattern library construction module is used for constructing a multi-element clustering pattern library according to the three-dimensional training data, the speed model and the density model; The pattern library screening module is used for constructing an alternative pattern library by utilizing the multi-element clustering pattern library according to the set simulation path and the space geological constraint condition; The model construction module is used for calculating a probability matrix according to the alternative mode library, and selecting modes according to the probability matrix and cross entropy to obtain a three-dimensional geological initial model; The model optimization module is used for performing multi-scale iterative optimization on the three-dimensional geological initial model by adopting the three-dimensional training data to obtain an optimized three-dimensional geological model; And constructing an alternative mode library by utilizing the multi-element clustering mode library according to the set simulation path and the space geological constraint condition, wherein the method comprises the following steps of: Sequentially determining the positions of nodes to be simulated according to the set simulation paths; Extracting a plurality of modes associated with the positions of the nodes to be simulated from the multi-element clustering mode library for each node to be simulated; screening each mode according to the space geological constraint condition to obtain a target mode; Obtaining an alternative mode library according to the target modes of the nodes to be simulated; the method comprises the steps of calculating a probability matrix according to the alternative mode library, selecting modes according to the probability matrix and cross entropy, and obtaining a three-dimensional geological initial model, and comprises the following steps: Calculating the lithology data similarity of each target mode in the candidate mode library through a Hamming distance function, and calculating the speed data similarity and the density data similarity of each target mode in the candidate mode library through a Euclidean distance function; obtaining a similarity matrix according to the lithology data similarity, the speed data similarity and the density data similarity; normalizing the data in the similarity matrix to obtain a probability matrix; calculating probability weight coefficients of the target modes by using cross entropy; carrying out weight adjustment on the probability matrix according to each probability weight coefficient to obtain a comprehensive probability matrix; and selecting a target mode with the highest probability value from the comprehensive probability matrix, and collaging the target mode to the node to be simulated to obtain a three-dimensional geological initial model.
7. A computer device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor implementing the method of constructing a three-dimensional geologic model of fused multi-data of any of claims 1-5 when the computer program is executed.
8. A computer readable storage medium, wherein the computer readable storage medium stores a computer program, and wherein the apparatus in which the computer readable storage medium is located implements the method for constructing a three-dimensional geological model by fusing multiple data according to any one of claims 1 to 5 when the computer program is executed.

Description

Method, device, equipment and medium for constructing three-dimensional geologic model fusing multiple metadata Technical Field The invention relates to the technical field of geological survey, in particular to a method, a device, equipment and a medium for constructing a three-dimensional geological model fusing multiple data. Background With the continuous development of geological exploration technology, three-dimensional geological modeling is increasingly widely applied to resource development and engineering investigation. However, the prior art still faces many challenges in constructing complex geologic structures. In the prior art, although the traditional multipoint statistics (MPS) method can simulate a geological structure, uncertainty and error of a model are large when data sparsity and geological constraints are insufficient. In addition, most of the existing MPS methods only extract pattern features from a single training image, and it is difficult to characterize global features of a complex geological structure, resulting in insufficient reliability and accuracy of simulation results. Meanwhile, the prior art lacks effective global feature extraction and geologic semantic consideration when fusing multi-element geological data, and is difficult to effectively reduce the uncertainty of a model. Therefore, how to effectively fuse multiple geological data in three-dimensional geological modeling to improve the accuracy of modeling has become a technical problem to be solved urgently by those skilled in the art. Disclosure of Invention The invention provides a three-dimensional geological model construction method, device, equipment and medium for fusing multiple data, so as to realize accurate model construction on a target geological region. In order to solve the above technical problems, an embodiment of the present invention provides a method for constructing a three-dimensional geologic model with fusion of metadata, including: and acquiring two-dimensional profile detection data of the target geological region, and expanding the two-dimensional profile detection data into three-dimensional training data. And extracting a space geological constraint condition from the two-dimensional section detection data. And constructing a multi-element clustering mode library according to the three-dimensional training data, the speed model and the density model. And constructing an alternative mode library by utilizing the multi-element clustering mode library according to the set simulation path and the space geological constraint condition. And calculating according to the alternative mode library to obtain a probability matrix, and selecting modes according to the probability matrix and cross entropy to obtain a three-dimensional geological initial model. And carrying out multi-scale iterative optimization on the three-dimensional geological initial model by adopting the three-dimensional training data to obtain an optimized three-dimensional geological model. Further, the two-dimensional profile detection data includes geological profile data, longitudinal wave velocity profile data, and density profile data. The expanding the two-dimensional profile detection data into three-dimensional training data comprises the following steps: And importing the geological profile data, the longitudinal wave speed profile data and the density profile data into a simulation grid, randomly expanding the simulation grid, and expanding the geological profile data, the longitudinal wave speed profile data and the density profile data into a three-dimensional profile with a certain width to obtain three-dimensional training data. Further, the extracting the spatial geological constraint condition from the two-dimensional section detection data includes: and extracting stratum sequence information and stratum thickness information from the geological profile data. And identifying and obtaining the space contact relation between the geological fault data and each geological body according to the geological profile data. And taking the stratum sequence information, stratum thickness information, geological fault data and spatial contact relation as the spatial geological constraint condition of the target geological region. Further, the constructing a multi-element clustering mode library according to the three-dimensional training data, the speed model and the density model comprises the following steps: and constructing a multi-element mode library according to the three-dimensional training data, the speed model and the density model. And carrying out cluster analysis on the modes of the multi-mode library according to a similarity threshold value, and classifying the multi-mode library according to an analysis result to obtain the multi-cluster mode library. Further, the constructing an alternative pattern library by using the multi-element clustering pattern library according to the set simulation path and the s