Identification of natural fractures in wellbore images using machine learning

Patent metadata

Field	Value
Publication number	US-10977489-B2
Application number	US-201816182818-A
Country	US
Kind code	B2
Filing date	Nov 7, 2018
Priority date	Nov 7, 2018
Publication date	Apr 13, 2021
Grant date	Apr 13, 2021

Abstract

A system, method and program product for processing borehole images to delineate between natural fractures and induced fractures. A system is disclosed that includes: an image analysis platform that inputs a noisy image from a borehole, processes the image using a set of filtering strategies, and renders a set of suggested filtered images via a user interface, the user interface including a mechanism for allowing a user to choose a selected filtered image from the set of suggested filtered images that best delineates between natural fractures and induced fractures, and wherein the image analysis platform further includes a feedback system for packaging and outputting the noisy image and selected filtered image as feedback; and a learning platform having a knowledge registration system that collects and stores training data and the feedback and in a knowledgebase, and a machine learning system that generates filtering strategies.

First claim

What is claimed is:

1. A fracture analysis system that processes borehole images to delineate between natural fractures and induced fractures, comprising:
an image analysis platform that inputs a noisy image from a borehole, processes the image using a set of filtering strategies, and renders a set of suggested filtered images via a user interface, the user interface including a mechanism for allowing a user to choose a selected filtered image from the set of suggested filtered images that best delineates between natural fractures and induced fractures, and wherein the image analysis platform further includes a feedback system for packaging and outputting the noisy image and selected filtered image as feedback; and
a learning platform having a knowledge registration system that collects and stores training data and the feedback and in a knowledgebase, and a machine learning system that generates filtering strategies that can be applied to noisy images to delineate between natural fractures and induced fractures based on the feedback and training data for different types of borehole images.

2. The fracture analysis system of claim 1 , wherein the machine learning system comprises a neural network that is trained with noisy/filtered image pairs obtained from the knowledgebase.

3. The fracture analysis system of claim 2 , wherein the neural network is further trained with a justification for choosing the selected filtered image.

4. The fracture analysis system of claim 2 , wherein the neural network is further trained based on a quality of each filtered image in a given noisy/filtered image pair.

5. The fracture analysis system of claim 1 , wherein the user interface further includes a mechanism for allowing the user to enter a justification for choosing the selected filtered image.

6. The fracture analysis system of claim 1 , wherein the borehole images are selected from a group consisting of: electrical imagery, acoustical imagery, optical imagery, video imagery or nuclear imagery.

7. The fracture analysis system of claim 1 , wherein the machine learning system utilizes one of: a Hough transformation and a pixel classification.

8. A computer program product for processing borehole images to delineate between natural fractures and induced fractures, the computer program product comprising a non-transitory computer readable storage medium having a set of instructions stored therein that, when executed by a processor, causes the processor to process borehole images by:
receiving a noisy image from a borehole;
processing the noisy image using a set of filtering strategies;
rendering a set of suggested filtered images via a user interface;
providing for a user to choose a selected filtered image from the set of suggested filtered images via the user interface that best delineates between natural fractures and induced fractures;
packaging and outputting the noisy image and selected filtered image as feedback;
collecting and storing training data and the feedback in a knowledgebase; and
generating filtering strategies that can be applied to noisy borehole images to delineate between natural fractures and induced fractures based on the feedback and training data for different types of borehole images.

9. The program product of claim 8 , wherein generating filtering strategies is performed by a neural network that is trained with noisy/filtered image pairs obtained from the knowledgebase.

10. The program product of claim 8 further causing the processor to process borehole images by providing for the user to enter a justification for choosing the selected filtered image.

11. The program product of claim 10 , wherein the neural network is further trained with the justification.

12. The program product of claim 8 , wherein the neural network is trained based on a quality of each filtered image in a given noisy/filtered image pair.

13. The program product of claim 8 , wherein the different types of borehole images include: electrical imagery, acoustical imagery, optical imagery, video imagery, and nuclear imagery.

14. The program product of claim 8 , wherein the filtering strategies utilize one of: a Hough transformation and a pixel classification.

15. A computerized method for processing borehole images to delineate between natural fractures and induced fractures, comprising:
inputting a noisy image from a borehole;
processing the noisy image using a set of filtering strategies;
rendering a set of suggested filtered images for selection via a user interface;
receiving from a user a selected filtered image from the set of suggested filtered images via the user interface that best delineates between natural fractures and induced fractures;
packaging and outputting the noisy image and selected filtered image as feedback;
collecting and storing training data and the feedback in a knowledgebase; and
generating the filtering strategies that can be applied to noisy images to delineate between natural fractures and induced fractures based on the feedback and training data for different types of borehole images;
wherein:
at least the processing, packaging and outputting, and generating steps are performed by computer software running on computer hardware.

16. The computerized method of claim 15 , wherein generating filtering strategies is performed by a neural network that is trained with noisy/filtered image pairs obtained from the knowledgebase.

17. The computerized method of claim 15 , further comprising providing for a user to enter a justification for choosing the selected filtered image.

18. The computerized method of claim 17 , wherein the neural network is further trained with the justification.

19. The computerized method of claim 15 , wherein the neural network is trained based on a quality of each filtered image in a given noisy/filtered image pair.

20. The computerized method of claim 15 , wherein the different types of borehole images include: electrical imagery, acoustical imagery, optical imagery; video imagery; and nuclear imagery.

Assignees

• IBM

Inventors

• Mena Paz Carmen N
• Kormaksson Matthias
• Cavalin Paulo R
• Borrel Paul

Classifications

• E21B47/002 Primary

  by visual inspection · CPC title

• G06V10/48

  by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation · CPC title

• G06V10/82

  using neural networks · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06N3/042

  Knowledge-based neural networks; Logical representations of neural networks · CPC title