CN-121995459-A - Fault and crack prediction method, device and computer equipment

CN121995459ACN 121995459 ACN121995459 ACN 121995459ACN-121995459-A

Abstract

The application provides a fault and crack prediction method and device and computer equipment, and belongs to the technical field of oil and gas exploration and development. The method comprises the steps of obtaining seismic data of a target area, identifying faults in the seismic data by utilizing a pre-built machine learning model to obtain a first fault data body, extracting seismic attribute data bodies sensitive to faults of different scales and seismic attribute data bodies sensitive to cracks from the seismic data, and carrying out multi-attribute fusion on the first fault data body, the seismic attribute data bodies sensitive to the faults of different scales and the seismic attribute data bodies sensitive to the cracks to identify the faults and the cracks of the target area of different scales. The method is based on the technical conception of machine learning algorithm and multi-seismic attribute fusion identification, and accurate prediction of faults and cracks with different scales is realized.

Inventors

ZHOU YAN
ZHANG SHIYUE
WU HAO
YAO GUANYU
YU XUE
ZHANG HONGJIE
LIU ZHIJUN
LI JIANHAI
Ji Luanming
SUN LONG

Assignees

中国石油天然气股份有限公司

Dates

Publication Date: 20260508
Application Date: 20241107

Claims (15)

1. A fault and fracture prediction method, comprising: acquiring seismic data of a target area; Identifying faults in the seismic data by utilizing a pre-constructed machine learning model to obtain a first fault data body; Extracting a seismic attribute data volume sensitive to faults of different scales and a seismic attribute data volume sensitive to cracks from the seismic data; and carrying out multi-attribute fusion on the first fault data body, the fault-sensitive seismic attribute data body with different scales and the crack-sensitive seismic attribute data body, and identifying faults and cracks with different scales of the target area.
2. The method as recited in claim 1, further comprising: Recovering the low-frequency and high-frequency seismic data to obtain the seismic data with double-side frequency expansion; Dividing frequency of the seismic data after bilateral frequency expansion, and respectively denoising different frequency segments obtained by the frequency division to obtain different frequency data bodies after denoising; Reconstructing the denoised data volumes with different frequencies to obtain reconstructed seismic data, and triggering and utilizing a pre-constructed machine learning model to identify faults in the seismic data to obtain a first fault data volume.
3. The method as recited in claim 1, further comprising: Enhancing a linear structure in the seismic attribute data volume sensitive to the crack to obtain a target area crack plan; The method comprises the steps of vectorizing a crack plane graph of a target area, and identifying the positions of line segments of each crack; And obtaining crack distribution characteristics of the target area according to the identified crack line segment positions, wherein the crack distribution characteristics comprise at least one of crack density characteristics and crack direction characteristics.
4. The method of claim 1, wherein extracting from the seismic data a volume of seismic attribute data sensitive to faults of different scales and a volume of seismic attribute data sensitive to fractures comprises: Extracting a curvature volume from the seismic data; And extracting a maximum likelihood volume from the seismic data.
5. The method of claim 4, wherein multi-attribute fusing the first fault data volume, the fault-sensitive seismic attribute data volume, and the fracture-sensitive seismic attribute data volume to identify faults and fractures of different scales for the target region, comprising: RGB fusion is carried out on the first fault data body, the curvature body and the maximum likelihood body, and a fusion data body is obtained; identifying faults of a first scale in the target area by using a first color data body representing a first fault data body in the fusion data body; identifying faults with a second scale in the target area by using a second color data body representing a curvature body in the fusion data body; Identifying a third-scale crack development zone associated with a crack in a crack network formed by a first-scale fault and a second-scale fault by using a third color data body representing the maximum likelihood in the fusion data body; Wherein the first dimension is greater than the second dimension, and the second dimension is greater than the third dimension.
6. The method of claim 1, wherein the machine learning model is a deep neural network model.
7. A method according to claim 3, wherein enhancing the linear structure in the fracture-sensitive seismic attribute data volume to obtain a target zone fracture plan comprises: generating a seismic profile using the fracture-sensitive seismic attribute data volume; constructing a Hessian matrix corresponding to the seismic section; Constructing Gaussian functions with different scale space factors, and respectively solving second-order partial derivatives of the abscissa and the ordinate of the seismic section by each Gaussian function; Convolving the second partial derivatives corresponding to the Gaussian functions with the Hessian matrixes respectively to obtain first matrixes, and identifying cracks and punctiform structures of the linear structures in the seismic section according to the eigenvalues and eigenvectors in the first matrixes; and constructing a response function according to the characteristic value and the characteristic vector, and utilizing the response function to strengthen the crack of the linear structure and filter the punctiform structure to obtain a target area crack plan.
8. The method of claim 7, wherein the enhancing the linear structure in the fracture-sensitive seismic attribute data volume to obtain the target zone fracture plan further comprises: And binarizing the cracks and the background in the target area crack plane graph to obtain a binarized target area crack plane graph.
9. The method of any one of claims 3, 7, 8, wherein the vectorizing the target area fracture plan to identify individual fracture line segment locations comprises: And carrying out Hough change on the target area crack plane graph, and identifying the crack line segment position.
10. The method of claim 4, wherein extracting curvature from the seismic data comprises: carrying out azimuth scanning on the seismic data to obtain seismic sub-data volumes with different azimuth; and respectively extracting curvature attributes of each seismic sub-data body, and obtaining each curvature body in one-to-one correspondence.
11. The method of claim 1, wherein the seismic data is post-stack narrow azimuth seismic data.
12. A fault and fracture prediction apparatus, comprising: the seismic data acquisition module is used for acquiring the seismic data of the target area; The first fault data body prediction module is used for identifying faults in the seismic data by utilizing a pre-constructed machine learning model to obtain a first fault data body; the seismic attribute extraction module is used for extracting a seismic attribute data volume sensitive to faults of different scales and a seismic attribute data volume sensitive to cracks from the seismic data; And the fault crack identification module is used for carrying out multi-attribute fusion on the first fault data body, the seismic attribute data body sensitive to faults of different scales and the seismic attribute data body sensitive to cracks, and identifying faults and cracks of different scales of the target area.
13. The apparatus of claim 12, wherein multi-attribute fusing the first volume of fault data, the volume of fault-sensitive seismic attribute data, and the volume of fault-sensitive seismic attribute data to identify faults and cracks of different scales for the target region comprises: RGB fusion is carried out on the first fault data body, the curvature body and the maximum likelihood body, and a fusion data body is obtained; identifying faults of a first scale in the target area by using a first color data body representing a first fault data body in the fusion data body; identifying faults with a second scale in the target area by using a second color data body representing a curvature body in the fusion data body; Identifying a third-scale crack development zone associated with a crack in a crack network formed by a first-scale fault and a second-scale fault by using a third color data body representing the maximum likelihood in the fusion data body; Wherein the first dimension is greater than the second dimension, and the second dimension is greater than the third dimension.
14. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the fault and fracture prediction method of any one of claims 1-11 when the program is executed.
15. A machine readable storage medium having stored thereon a computer program, which when executed by a processor implements the fault and fracture prediction method of any one of claims 1 to 11.

Description

Fault and crack prediction method, device and computer equipment Technical Field The application belongs to the technical field of oil and gas exploration and development, and particularly relates to a fault and crack prediction method, a fault and crack prediction device, computer equipment and a machine-readable storage medium. Background Shallow layer fracture development in southern pine, fracture and associated cracks play a vital role in oil and gas reservoir and enrichment. In the unconventional oil reservoir, the fracture and the associated cracks increase the porosity of a matrix on one hand, and on the other hand, the artificial cracks generated by fracturing communicate massive natural cracks, a seepage system is modified, and the yield is improved. Through multiple development area anatomies, it is believed that the closer to fault, the higher the cumulative yield, the more the formation shale oil demonstrates that the fracture zone yield is significantly better than the non-fracture zone. In horizontal well drilling, fracture and crack development can increase the risks of borehole collapse, leakage and casing deformation, and influence whether the horizontal well can be normally drilled and successfully fractured, and therefore, early warning and pre-judging are needed for micro-fracture and crack development before and during horizontal well drilling. Through forward analysis, faults with vertical fault distances larger than 10m can be identified when the main frequency of the earthquake is 40Hz under ideal conditions, and faults with vertical fault distances about 5m can be identified when the main frequency of the earthquake is 80 Hz. At present, the actual main frequency of the earthquake is 40-50 Hz, and the faults with the minimum vertical fault distance of 15-20 m on the earthquake can be directly identified by naked eyes under the influence of signal-to-noise ratio and other factors. At present, by utilizing the difference of fault and crack sensitivity of different seismic attributes, a plurality of seismic fracture prediction schemes are proposed in the industry, for example, by utilizing the characteristic that a coherent body is sensitive to faults and the characteristic that a maximum likelihood body is sensitive to crack bands, the coherent body and the maximum likelihood body are extracted from post-stack seismic data, the faults with the vertical fault distance of 15-20 m are identified by utilizing the extracted coherent body, and the crack development is identified by utilizing the maximum likelihood body, however, the fault response in the seismic data can be enhanced by the coherent body, but the identification result is influenced by noise and other similar fault responses, the extracted fault surface is smaller, the combination is difficult, and the fault prediction precision is lower. Accordingly, there is a need for improvements in the above-described seismic fracture prediction schemes. Disclosure of Invention The embodiment of the application aims to provide a fault and crack prediction method, a fault and crack prediction device, computer equipment and a machine-readable storage medium, which are used for solving the technical problems that in the prior art, the accuracy of a prediction result obtained by using a coherent body for fracture prediction is low, the fault level is small and the combination is difficult. To achieve the above object, a first aspect of the present application provides a fault and fracture prediction method, the method comprising: acquiring seismic data of a target area; Identifying faults in the seismic data by utilizing a pre-constructed machine learning model to obtain a first fault data body; Extracting a seismic attribute data volume sensitive to faults of different scales and a seismic attribute data volume sensitive to cracks from the seismic data; and carrying out multi-attribute fusion on the first fault data body, the fault-sensitive seismic attribute data body with different scales and the crack-sensitive seismic attribute data body, and identifying faults and cracks with different scales of the target area. In a specific embodiment of the present application, the method further includes: Recovering the low-frequency and high-frequency seismic data to obtain the seismic data with double-side frequency expansion; Dividing frequency of the seismic data after bilateral frequency expansion, and respectively denoising different frequency segments obtained by the frequency division to obtain different frequency data bodies after denoising; Reconstructing the denoised data volumes with different frequencies to obtain reconstructed seismic data, and triggering and utilizing a pre-constructed machine learning model to identify faults in the seismic data to obtain a first fault data volume. In a specific embodiment of the present application, the method further includes: Enhancing a linear structure in the seismic attribute data volume sens