Artifact removal from borehole images using machine learning

Patent metadata

Field	Value
Publication number	US-12299846-B2
Application number	US-202217678832-A
Country	US
Kind code	B2
Filing date	Feb 23, 2022
Priority date	Jan 31, 2022
Publication date	May 13, 2025
Grant date	May 13, 2025

Abstract

A method for correcting borehole images may include acquiring a raw data image of a formation using a downhole tool that takes one or more measurements and processing the raw data image through a machine learning model to form a corrected image. The method may further include displaying the corrected image and identifying one or more formation properties based at least in part on the corrected image.

First claim

What is claimed is:

1. A method for correcting borehole images comprising:
acquiring a raw data image of a formation using a downhole tool that takes one or more measurements;
processing the raw data image through a machine learning model to form a corrected image, wherein the machine learning model is determined using a training dataset, wherein one or more inputs of the training dataset comprise an image degrading pattern and azimuthal coverage of a borehole in a subterranean formation;
displaying the corrected image; and
identifying one or more formation properties based at least in part on the corrected image.

2. The method of claim 1 , wherein the downhole tool is a resistivity imager, and the raw data image represents a resistivity of the formation.

3. The method of claim 1 , wherein the downhole tool is an impedance imager, and the raw data image represents an impedivity of the formation.

4. The method of claim 3 , wherein the downhole tool operates at one or more frequencies, and the machine learning model is used to correct the raw data image for each of the one or more frequencies.

5. The method of claim 3 , wherein the downhole tool operates at one or more frequencies, and a second machine learning model is used to correct the raw data image for each of the one or more frequencies.

6. The method of claim 3 , wherein the one or more measurements are one or more complex-numbers and the raw data image represent a real and an imaginary component of the one or more measurements, and a second machine learning model is used to correct the raw data image.

7. The method of claim 1 , wherein the machine learning model removes one or more artifacts in the raw data image.

8. The method of claim 1 , wherein the machine learning model is trained with the raw data image and the corrected image that were processed using rule-based corrections or expert-based corrections.

9. The method of claim 1 , wherein the machine learning model is trained using the raw data image or the corrected image from a section of a log with high image quality.

10. The method of claim 1 , wherein the machine learning model is trained using the raw data image with a noise or an imposed artifact, or the corrected image with the noise or the imposed artifact.

11. The method of claim 1 , wherein processing the raw data image through the machine learning model occurs at a plurality of depths and the corrected image for each of the plurality of depths is displayed in real-time.

12. The method of claim 1 , wherein processing the raw data image through the machine learning model occurs over a window of data of preset size and the corrected image is displayed in near real-time.

13. The method of claim 1 , further comprising applying a blending algorithm to the corrected image to correct for a dielectric effect and a standoff effect.

14. The method of claim 1 , further comprising adjusting a logging speed, a pad pressure, an operating frequency, or a power level based at least in part on the corrected image.

15. The method of claim 1 , further comprising applying an inversion algorithm to the corrected image to correct for a dielectric effect and a standoff effect.

16. The method of claim 1 , further comprising identifying a geology of the formation based at least in part on the corrected image.

17. The method of claim 1 , further comprising processing the corrected image with a second machine learning model to form a blended image that combines a plurality of corrected images.

18. The method of claim 1 , further comprising training the machine learning model using a plurality of real and imaginary raw data images at one or more frequencies as inputs and a plurality of blended images as an output, wherein the machine learning models forms the corrected image that is corrected for one or more artifacts, a dielectric effect, and a standoff effect.

19. The method of claim 1 , wherein the machine learning model is a neural network or a decision-tree.

20. A system comprising:
a borehole imager tool comprising:
an injector electrode configured to inject current into a borehole; and
a return electrode configured to receive current from the borehole; and

an information handling system configured to:
acquire a raw data image of a formation using a downhole tool that takes one or more measurements, wherein the raw data image is determined using a training dataset with known inputs and outputs, wherein the inputs of the training dataset are based in part on added noise patterns and azimuthal coverage of a borehole in a subterranean formation;

process the raw data image through a regression function using the training dataset to form a corrected image;
display the corrected image; and
identify one or more formation properties based at least in part on the corrected image.

Assignees

• Halliburton Energy Services Inc

Inventors

• Guner Baris
• Fouda Ahmed E

Classifications

• G06T5/77

  Retouching; Inpainting; Scratch removal · CPC title

• E21B47/0025 Primary

  generating an image of the borehole wall using down-hole measurements, e.g. acoustic or electric · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T5/60

  using machine learning, e.g. neural networks · CPC title