CN-122023136-A - Automatic fault detection method based on multi-feature fusion and construction guide filtering

CN122023136ACN 122023136 ACN122023136 ACN 122023136ACN-122023136-A

Abstract

The application discloses a fault automatic detection method based on multi-feature fusion and structure-oriented filtering, which relates to the technical field of geological data detection and comprises the steps of obtaining a three-dimensional coherent body corresponding to seismic data of a target area and reflecting stratum discontinuity; the method comprises the steps of processing fault characteristics in a three-dimensional coherent body through a multi-scale multi-direction filter to generate a fault boundary enhanced image, calculating to obtain direction diffusion coefficients corresponding to a main direction gradient and a vertical direction gradient according to a fault trend angle in the three-dimensional coherent body, performing iterative diffusion filtering processing on the three-dimensional coherent body based on the direction diffusion coefficients to generate a fault continuity enhanced image, and generating a fault detection result based on a fusion image between the fault boundary enhanced image and the fault continuity enhanced image. The application reduces the false detection rate in the broken layer identification process and improves the identification precision.

Inventors

LIANG YU
YANG CHUNSHENG
ZHAO YUNFEI
ZHONG ZHISEN
YANG HUIDONG
CAI DONGMEI
Fu Xiandi
SHAN GAOJUN
LI HAO
LI HONGXING

Assignees

大庆油田有限责任公司
中国石油天然气股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260202

Claims (10)

1. A fault automatic detection method based on multi-feature fusion and construction-oriented filtering, the method comprising: acquiring a three-dimensional coherent body corresponding to seismic data of a target area and reflecting formation discontinuity; processing the fault characteristics in the three-dimensional coherent body through a multi-scale multi-directional filter to generate a fault boundary enhanced image; according to the fault trend angle in the three-dimensional coherent body, calculating to obtain a main direction gradient and a direction diffusion coefficient corresponding to the vertical direction gradient; performing iterative diffusion filtering processing on the three-dimensional coherent body based on the direction diffusion coefficient to generate a fault continuity enhanced image; and generating a fault detection result based on a fusion image between the fault boundary enhanced image and the fault continuity enhanced image.
2. The method for automatically detecting faults based on multi-feature fusion and structure-oriented filtering according to claim 1, wherein the processing fault features in the three-dimensional coherent body through a multi-scale multi-directional filter to generate a fault boundary enhanced image comprises: generating a tomographic image based on a tomographic directional feature in the three-dimensional coherent body; Carrying out convolution processing on the tomographic image through a multi-scale multi-direction filter to obtain a first enhanced image; and carrying out contrast analysis processing on the pixel values in the first enhanced image to obtain a fault boundary enhanced image.
3. The multi-feature fusion and construction-oriented filtering-based tomographic automatic detection method according to claim 2, wherein the generating a tomographic image based on tomographic direction features in the three-dimensional coherent volume includes: intercepting two-dimensional plane data in the three-dimensional coherent body according to the time depth direction to obtain coherent body slice data; Calculating the gradient and the slope direction of each pixel point based on the horizontal coordinate and the vertical coordinate of each pixel point in the coherent body slice data, wherein the gradient and the slope direction are used for representing the fault direction characteristics in the three-dimensional coherent body; Calculating to obtain a first pixel value of each pixel point after fault enhancement based on the gradient and the gradient direction; and updating each pixel point based on the first pixel value to generate a tomographic image.
4. The automatic fault detection method based on multi-feature fusion and structure-oriented filtering according to claim 2, wherein the convolution processing is performed on the fault image by a multi-scale multi-directional filter to obtain a first enhanced image, comprising: determining a plurality of division scales and division directions of the multi-scale multi-direction filter; based on different division scales and different division directions, carrying out convolution processing on the tomographic image through a multi-scale multi-direction filter; taking the maximum value of convolution results of each pixel point under different scales and different directions as a filtering processing result of each pixel point; And generating a first enhanced image based on the filtering processing result.
5. The automatic fault detection method based on multi-feature fusion and structure-oriented filtering as claimed in claim 2, wherein said performing a contrast analysis process on pixel values in the first enhanced image to obtain a fault boundary enhanced image includes: Determining a global statistical value in the first enhanced image and a neighborhood median value of each pixel point, wherein the global statistical value comprises a pixel value median value, a preset pixel high threshold value and a preset pixel low threshold value; Based on the neighborhood median value, a preset pixel high threshold value and a preset pixel low threshold value, carrying out contrast analysis processing on pixel values of all the pixel points, and determining fault types of all the pixel points, wherein the fault types comprise strong faults, undetermined layers and non-faults; Taking a pixel point in the strong fault as a seed point, and judging connectivity of neighborhood pixels of the seed point through a region growing algorithm to obtain a fault pixel set in a preset growing range; and generating a fault boundary enhanced image based on the fault pixel set.
6. The automatic fault detection method based on multi-feature fusion and construction guide filtering according to claim 5, wherein the comparing and analyzing the pixel value of each pixel point based on the neighborhood median, the preset pixel high threshold and the preset pixel low threshold to determine the fault type of each pixel point comprises: if the pixel value of the pixel point is larger than or equal to a preset pixel high threshold value, dividing the pixel point into strong faults; If the pixel value of the pixel point is larger than or equal to a preset pixel low threshold value, smaller than a preset pixel high threshold value and larger than the neighborhood median value, dividing the pixel point into a layer to be determined; And if the pixel value of the pixel point is smaller than or equal to the neighborhood median value or smaller than a preset pixel low threshold value, dividing the pixel point into non-faults.
7. The automatic fault detection method based on multi-feature fusion and construction guide filtering according to claim 1, wherein the direction diffusion coefficients include a main direction diffusion coefficient and a vertical direction diffusion coefficient, the direction diffusion coefficients corresponding to both the main direction gradient and the vertical direction gradient are calculated according to the fault trend angle in the three-dimensional coherent body, and the method comprises the following steps: Constructing a main direction gradient operator and a vertical direction gradient operator based on fault trend angles in the three-dimensional coherent body; The main direction gradient operator and the vertical direction gradient operator are respectively convolved with coherent volume slice data to obtain main direction gradient and vertical direction gradient, the coherent volume slice data are obtained after the three-dimensional coherent volume is intercepted, the main direction gradient is used for representing the edge continuity characteristic of a fault along a trend direction, and the vertical direction gradient is used for representing noise or non-fault interference signals perpendicular to the trend of the fault; And constructing a main direction diffusion coefficient based on the main direction gradient, and taking a preset fixed diffusion coefficient as a vertical direction diffusion coefficient corresponding to the vertical direction gradient.
8. The method for automatically detecting faults based on multi-feature fusion and structure-oriented filtering as claimed in claim 7, wherein the step of performing iterative diffusion filtering processing on the three-dimensional coherent body based on the direction diffusion coefficient to generate fault continuity enhanced images comprises the steps of: Determining a time step in the iterative diffusion process; And performing iterative diffusion filtering processing on the three-dimensional coherent body based on the time step, the main direction gradient amplitude, the vertical direction gradient amplitude, the main direction diffusion coefficient and the vertical direction diffusion coefficient to obtain a fault continuity enhanced image.
9. The multi-feature fusion and construction-oriented filtering-based fault automatic detection method according to claim 1, wherein the generating a fault detection result based on a fusion image between the fault boundary enhanced image and the fault continuity enhanced image comprises: Fusing the fault boundary enhanced image and the fault continuity enhanced image to obtain a fused image; and connecting discontinuous faults in the fused image based on a preset morphological algorithm to generate a fault detection result.
10. The automatic fault detection method based on multi-feature fusion and structure-oriented filtering as claimed in claim 9, wherein the connecting the discontinuous faults inside the fused image based on a preset morphological algorithm to generate a fault detection result comprises: Based on a preset morphological algorithm, performing closed operation on discontinuous faults in the fused image to obtain a connected fault zone; performing edge pixel stripping treatment on the connected fault zone to obtain a fault trend line; Converting the fault trend line into a discrete coordinate point set, and taking the discrete coordinate point set as a fault detection result, wherein the fault detection result comprises a fault position, a fault length and fault trend information.

Description

Automatic fault detection method based on multi-feature fusion and construction guide filtering Technical Field The application relates to the technical field of geological data detection, in particular to a fault automatic detection method based on multi-feature fusion and structure-oriented filtering. Background The fault is a structure that the stratum is broken under the stress action and the stratum on two sides of the fault is obviously displaced along the fracture surface, fault identification research is carried out in each stage of oil and gas field exploration, evaluation and development, and accurate fault identification has important guiding significance for reasonable division of development units, adjustment and optimization of well distribution schemes, reduction of development risks and the like. At present, fault identification based on seismic data is the basis of fault research and mainly comprises the following steps of determining the position, trend and breaking distance of faults based on a seismic reflection characteristic analysis method by analyzing the characteristics of dislocation, waveform distortion, amplitude mutation and the like of a phase axis of seismic reflection waves, carrying out wavelet transformation on the seismic data based on a frequency division technology, extracting a fault sensitive frequency band single-frequency data body, analyzing the amplitude abnormal characteristics of the fault sensitive frequency band single-frequency data body and carrying out fault identification. Disclosure of Invention In order to solve the technical problems of high false detection rate and low recognition accuracy in fault recognition caused by noise interference and discontinuous fracture in the recognition process and not considering data characteristics in the scenes, the application provides a fault automatic detection method based on multi-feature fusion and structure-oriented filtering. The adopted technical scheme is as follows: acquiring a three-dimensional coherent body corresponding to seismic data of a target area and reflecting formation discontinuity; processing the fault characteristics in the three-dimensional coherent body through a multi-scale multi-directional filter to generate a fault boundary enhanced image; According to the fault trend angle in the three-dimensional coherent body, calculating to obtain a main direction gradient and a direction diffusion coefficient corresponding to the vertical direction gradient; Performing iterative diffusion filtering processing on the three-dimensional coherent body based on the directional diffusion coefficient to generate a fault continuity enhanced image; and generating a fault detection result based on a fusion image between the fault boundary enhanced image and the fault continuity enhanced image. In one possible embodiment of the present application, processing a tomographic feature in a three-dimensional coherent volume by a multi-scale multi-directional filter to generate a tomographic boundary enhancement image includes: Generating a tomographic image based on the tomographic directional features in the three-dimensional coherent body; Convolving the tomographic image through a multi-scale multi-directional filter to obtain a first enhanced image; and carrying out contrast analysis processing on pixel values in the first enhanced image to obtain a fault boundary enhanced image. In one possible embodiment of the present application, generating a tomographic image based on a tomographic direction feature in a three-dimensional coherent volume includes: Intercepting two-dimensional plane data in a three-dimensional coherent body according to the time depth direction to obtain coherent body slice data; Calculating the gradient and the slope direction of each pixel point based on the horizontal coordinate and the vertical coordinate of each pixel point in the coherent body slice data, wherein the gradient and the slope direction are used for representing the fault direction characteristics in the three-dimensional coherent body; Calculating to obtain a first pixel value of each pixel point after fault enhancement based on the gradient and the gradient direction; and updating each pixel point based on the first pixel value to generate a tomographic image. In one possible embodiment of the present application, the convolution processing is performed on the tomographic image by a multi-scale multi-directional filter to obtain a first enhanced image, including: determining a plurality of division scales and division directions of the multi-scale multi-direction filter; Based on different division scales and division directions, carrying out convolution processing on the tomographic image through a multi-scale multi-direction filter; taking the maximum value of convolution results of each pixel point under different scales and different directions as a filtering processing result of each pixel point; based on the filter