CN115272156B - High resolution wellbore imaging characterization method for oil and gas reservoirs based on cyclic generative adversarial network - Google Patents

Abstract

The invention discloses a high-resolution shaft imaging characterization method of a hydrocarbon reservoir based on a cyclic generation countermeasure network, which relates to the technical field of geological exploration and development evaluation of hydrocarbon resources and comprises the following steps: s1, preprocessing a resistivity imaging image; s2, preprocessing the image of the outer surface of the full-diameter rock core; s3, mapping the resistivity imaging image to a full-diameter rock core outer surface image; s4, imaging and characterizing the high-resolution shaft of the oil and gas reservoir. The method realizes the imaging characterization of the high-resolution shaft of the oil and gas reservoir, effectively fuses the advantages of logging data and petrophysical data, overcomes the defect of representativeness deficiency of petrophysical experiments, improves the longitudinal resolution of logging images by at least 10 times, improves the dividing capability of thin hydrocarbon reservoir layers, has low cost of the plain scanning experiment of the cylindrical rock core on the outer surface, reduces the cost of the imaging characterization of the high-resolution shaft, is easy to apply and popularize in different oil and gas fields, and lays a foundation for the fine evaluation of the petrophysical parameters of the oil and gas reservoir and the construction of the three-dimensional digital shaft.

Description

High resolution wellbore imaging characterization method for oil and gas reservoirs based on cyclic generative adversarial network

technical field

The invention relates to the technical field of geological exploration and development evaluation of oil and gas resources, in particular to a high-resolution wellbore imaging representation method for oil and gas reservoirs based on cyclically generated confrontation networks.

Background technique

High-precision formation characterization has always been a difficult point in the fine exploration and development of oil and gas reservoirs at home and abroad. Although the current micro-nano-scale rock physics experiments, such as CT technology, scanning electron microscope technology, laser confocal technology, etc., can better characterize the characteristics of the mineral components and pore structure inside the rock, but the cost is high and the experiment period is long. Moreover, it is not representative for formations with strong heterogeneity, and it is difficult to carry out massive petrophysical experiments for each well. Logging technology can continuously collect formation physical field parameters for the entire well, but the resolution is low. At present, even the resistivity imaging logging image with the highest vertical resolution has a resolution of only 2.0mm. For thin layers with high oil and gas Effective identification cannot be carried out, and the evaluation accuracy is lower than the results of petrophysical experiments.

Deep learning technology is a good technology to solve the nonlinear physical response relationship. In addition to being applied to the construction of multi-scale digital core images and automatic segmentation of mineral components in micro-nano scale rock physics experiments, it can also be applied to well logging data. Realize different lithofacies division, porosity calculation, etc. However, so far, there have been no relevant achievements in the application of deep learning to effectively integrate logging technology and micro-nano scale petrophysical experiments. In order to fully combine the advantages of large-scale logging and high-precision petrophysical experiments at the micro-nano scale, it is necessary to use deep learning to achieve high-resolution wellbore imaging characterization of oil and gas reservoirs. Conventional adversarial network models require a one-to-one correspondence between resistivity imaging images and high-resolution petrophysical experimental images, which is strictly required, and because the drilling tool has a certain thickness, the pore structure characteristics cannot be unified. The cyclic generation confrontation network algorithm developed in recent years is not limited by the corresponding relationship, and can be used as a method for high-resolution wellbore imaging characterization of oil and gas reservoirs, reducing the cost of high-precision formation characterization, overcoming the lack of low vertical resolution of logging technology, and improving oil and gas reservoirs. The ability to divide thin layers has laid a foundation for the fine evaluation of petrophysical parameters of oil and gas reservoirs and the construction of 3D digital wellbores.

Contents of the invention

In order to solve the above technical problems, the present invention discloses a high-resolution wellbore imaging characterization method for oil and gas reservoirs based on cyclically generated confrontation networks.

To achieve the above object, the present invention adopts the following technical solutions:

A high-resolution wellbore imaging characterization method for oil and gas reservoirs based on cyclic generative confrontation network, including the following steps:

S1, resistivity imaging image preprocessing;

S2, image preprocessing of the outer surface of the full-diameter core;

S3, the mapping of the resistivity imaging image to the outer surface image of the full-diameter core;

S4, high-resolution wellbore imaging characterization of oil and gas reservoirs.

Optionally, in S1, the step of preprocessing the resistivity imaging image includes:

Collect the data collected by the resistivity imaging logging tool, generate resistivity imaging dynamic images through calibration and image enhancement methods, and then combine statistical algorithms to fill the blank strips of resistivity imaging dynamic images, remove resistivity abnormal values, and replace As the average value in the range A×A around this point, the resistivity is converted into porosity by using the Simeon’s formula of the shaly rock model, and the resistivity imaging image representing the porosity is obtained, and converted into [0,255] 8-bit resistivity imaging grayscale image with interval distribution.

Optionally, in S2, the step of preprocessing the image of the outer surface of the full-diameter rock core includes:

Collect the full-diameter cores collected by drilling, use the charge-coupled device in the core scanning analyzer to scan the cylindrical cores on the 360° outer surface with ordinary light, and obtain images with micron-level resolution, and distinguish pores and clay according to the shape and geometric characteristics of rock components Minerals and different framework components, and converted to an 8-bit grayscale image of the outer surface of the full-diameter core distributed in the [0,255] interval.

Optionally, in S3, the step of mapping the resistivity imaging image to the outer surface image of the full-diameter core includes:

S3.1: Divide the resistivity imaging grayscale image obtained in S1 into K resistivity imaging grayscale images with M×M pixels, and compress the grayscale image of the outer surface of the full-diameter core obtained in S2 to the resolution of the resistivity imaging image Q times of the rate, and divided into K full-diameter rock core outer surface grayscale images with (M×Q)×(M×Q) pixels, where Q is an integer;

S3.2: Take the K grayscale images of resistivity imaging after S3.1 segmentation as the X domain, and the segmented K grayscale images of the outer surface of the full-diameter core as the Y domain, and then use the loop to generate the confrontation loss of the confrontation network algorithm The function learns the mapping function from the X domain to the Y domain, and realizes the mapping from the resistivity imaging image to the outer surface image of the full-diameter core.

Optionally, in S4, the step of high-resolution wellbore imaging and characterization of oil and gas reservoirs includes:

Compare the image effects generated under different training times in S3.2, optimize the number of loop iterations, apply the optimal mapping model to the K resistivity imaging grayscale images obtained in S3.1, and finally process the images according to the sampling depth Stitching to realize high-resolution wellbore imaging characterization of oil and gas reservoirs.

The beneficial effect of the present invention is that the method realizes the high-resolution wellbore imaging representation of oil and gas reservoirs, effectively integrates the advantages of well logging data and petrophysical data, makes up for the problem of insufficient representativeness of petrophysical experiments, and makes the vertical resolution of well logging images The rate has been increased by at least 10 times, which improves the division ability of thin layers of oil and gas layers, and the cost of ordinary light scanning experiments on cylindrical cores on the outer surface is low, which reduces the cost of high-resolution wellbore imaging and characterization, and is easy to apply and popularize in different oil and gas fields. The fine evaluation of physical parameters and the construction of 3D digital wellbore laid the foundation.

Description of drawings

Fig. 1 is a flow chart of a high-resolution wellbore imaging characterization method for oil and gas reservoirs based on cyclic generation confrontation network of the present invention;

Fig. 2 is the comparison chart of resistivity imaging image and processed image of the present invention, (a) is the original dynamic resistivity imaging image of well A 4621.9m-4623.6m depth section, (b) is the 8-bit resistivity after the blank strip is filled Imaging a grayscale image, (c) is a synthetic high-resolution image.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

Taking Well A of an oilfield in China as an example, the present invention is based on the cycle-generated confrontation network high-resolution wellbore imaging representation method for oil and gas reservoirs, as shown in Figure 1, including the following steps:

S1, Resistivity Imaging Image Preprocessing

The data collected by the resistivity imaging logging tool in Well A is collected, and the original dynamic resistivity imaging image is generated through velocity correction, abnormal electrode correction, voltage correction, swing arm correction and image enhancement methods, as shown in Fig. 2(a) , combined with the multi-point statistical algorithm to fill the blank strips of the dynamic image of the resistivity imaging, remove the abnormal value of the resistivity, and replace it with the average value within the range of 3×3 around the abnormal point, and use the gamma logging to calculate the formation shale content, and then convert the resistivity into porosity by using the Simon's formula of the shale-bearing rock model, obtain the resistivity imaging image representing the porosity, and convert it into an 8-bit resistivity imaging grayscale image distributed in the [0,255] interval , as shown in Figure 2(b);

S2, image preprocessing of the outer surface of the full-diameter core

Collect the full-diameter cores of Well A collected by drilling, use the charge-coupled device (CCD) in the core scanning analyzer to scan the cylindrical cores on the outer surface of 360° in general light, and obtain the ordinary light scanning images with micron resolution. The geometric characteristics of component shapes distinguish pores, clay minerals and different framework components, and convert them into 8-bit grayscale images of the outer surface of the full-diameter core distributed in the [0,255] interval;

S3, Mapping of resistivity imaging image to the outer surface image of the full-diameter core

S3.1: Segmentation of the resistivity imaging image and the outer surface image of the full-diameter core;

The resistivity imaging grayscale image obtained by S1 is divided into 7400 images with a size of 100×100 pixels (resolution 2.0mm), and the grayscale image of the outer surface of the full-diameter core obtained by S2 (resolution 0.021mm) is compressed to 16 times the resolution of the resistivity imaging image (resolution 0.125mm), and divided into 7400 images with 1600×1600 pixels;

S3.2: Establish the mapping relationship from the resistivity imaging image to the outer surface image of the full-diameter core;

The 7,400 grayscale images of electrical resistivity imaging after S3.1 segmentation are used as the X domain, and the 7,400 grayscale images of the outer surface of the full-diameter core after segmentation are used as the Y domain, and then the X domain is learned by using the adversarial loss function of the cyclic generative adversarial network algorithm The mapping function to the Y domain establishes the mapping relationship from the resistivity imaging image to the outer surface image of the full-diameter core;

S4, high-resolution wellbore imaging characterization of oil and gas reservoirs

Comparing the image effects generated under different training times in S3.2, it is determined that the optimal number of loop iterations is 40, and the optimal mapping model is applied to the 7400 resistivity imaging grayscale images obtained in S3.1, as shown in the figure As shown in 2(c), the images are finally stitched together according to the sampling depth to achieve high-resolution wellbore imaging characterization of oil and gas reservoirs.

Of course, the above descriptions are not intended to limit the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or replacements made by those skilled in the art within the scope of the present invention shall also belong to the present invention. protection scope of the invention.

Claims (1)

1. A high resolution wellbore imaging characterization method for a hydrocarbon reservoir based on a cyclical generation antagonism network, comprising the steps of:

s1, preprocessing a resistivity imaging image;

s2, preprocessing the image of the outer surface of the full-diameter rock core;

s3, mapping the resistivity imaging image to a full-diameter rock core outer surface image;

s4, imaging and characterizing the high-resolution shaft of the oil and gas reservoir;

in S1, the step of preprocessing the resistivity imaging image includes:

collecting data acquired by a resistivity imaging logging instrument, generating a resistivity imaging dynamic image through a correction and image enhancement method, filling blank strips of the resistivity imaging dynamic image by combining a statistical algorithm, removing abnormal resistivity values, and replacing the blank strips with surrounding pointsA×AAverage value in range, then using Siemens formula of shale-containing rock model to convert resistivity into porosity to obtain resistivity imaging image of representing porosity, and converting it into [0,255]]8-bit resistivity imaging gray scale images distributed in intervals;

s2, preprocessing the image of the outer surface of the full-diameter rock core, wherein the preprocessing comprises the following steps:

collecting a full-diameter rock core acquired by drilling, performing 360-degree outer surface cylindrical rock core plain scanning by using a charge coupled device in a rock core scanning analyzer to obtain an image with micron-scale resolution, distinguishing pores, clay minerals and different framework components according to geometric characteristics of the shape of rock components, and converting the image into 8-bit full-diameter rock core outer surface gray images distributed in the interval of [0,255 ];

in S3, the mapping of the resistivity imaging image to the full diameter core outer surface image includes:

s3.1: dividing the resistivity imaging gray scale image obtained in the step S1 into the image with the following stepsM×MA pixel ofKZhang Dianzu imaging gray level image, compressing the gray level image of the outer surface of the full-diameter rock core obtained in S2 to the resolution of the resistivity imaging imageQDoubling and dividing intoM×Q）×（M×Q) A pixel ofKZhang Quan diameter core outer surface grayscale image, wherein,Qis an integer;

s3.2: dividing S3.1KZhang Dianzu Rate imaging Gray image asXDomain, segmentedKZhang Quan diameter core outer surface gray scale image asYDomain, reuse loop generates countermeasures loss function learning for countermeasures network algorithmsXDomain to domainYMapping function of domain, realizing mapping from resistivity imaging image to full diameter core outer surface image;

s4, the step of imaging and characterizing the high-resolution shaft of the oil and gas reservoir comprises the following steps:

comparing the image effects generated under different training times in S3.2, optimizing the cycle iteration times, and applying the optimal mapping model to the image obtained in S3.1KAnd in the Zhang Dianzu-rate imaging gray level images, finally splicing the images according to the sampling depth to realize the imaging characterization of the high-resolution shaft of the oil and gas reservoir.

Images