Back to results
US 12,710,414 B2Granted
System and method to accelerate core images acquisition and processing using machine learning
AssigneeSaudi Arabian Oil Company
InventorsTao Lin, Weichang Li, Mokhles M. Mezghani
Life of this patent
Estimating the term…
Abstract
A method for analyzing rock cores of a subterranean formation is disclosed. The method includes capturing low resolution core images of the rock cores, selecting, by a computer processor and based on a pre-determined quality threshold for qualifying the low resolution core images, a number of qualified rock cores, capturing high resolution core images of the qualified rock cores, generating, by the computer processor and based on a high resolution core image evaluation model, a ranking of the qualified rock cores, and analyzing, based at least on the ranking, the qualified rock cores to generate a core analysis result.
Claims11
A method comprising:
- capturing a training dataset comprising a plurality of coarse core images of a plurality of rock cores and a plurality of rock core labels associated with the plurality of coarse core images;
- performing a training operation on a convolutional neural network using the training dataset and a regression algorithm to produce a trained model,
- wherein the trained model determines one or more predicted image scores in response to receiving one or more core images, and
- wherein the one or more predicted image scores are used to determine a selection among a plurality of image processing operations;
- determining, by a computer processor, a predicted image score using the trained model and a coarse core image of a rock core;
- determining, by the computer processor, a selected image processing operation among the plurality of image processing operations based on the predicted image score,
- wherein the selected image processing operation corresponds to an image acquisition of the rock core at a higher resolution than the coarse core image;
- and
- capturing, by a camera and in response to determining the selected image processing operation, a higher resolution core image of the rock core;
- and
- determining, by the computer processor, a permeability or a porosity of the rock core using the higher resolution core image.
8 more elements
A core image analyzer, comprising:
- a processor;
- and
- a memory coupled to the processor and storing instructions, the instructions, when executed by the processor, being configured to perform a method comprising:
- capturing a training dataset comprising a plurality of coarse core images of a plurality of rock cores and a plurality of rock core labels associated with the plurality of coarse core images;
- performing a training operation on a convolutional neural network using the training dataset and a regression algorithm to produce a trained model,
- wherein the trained model determines one or more predicted image scores in response to receiving one or more core images, and
- wherein the one or more predicted image scores are used to determine a selection among a plurality of image processing operations;
- determining a predicted image score using the trained model and a coarse core image of a rock core;
- determining a selected image processing operation among the plurality of image processing operations based on the predicted image score,
- wherein the selected image processing operation corresponds to an image acquisition of the rock core at a higher resolution than the coarse core image;
- and
- capturing, by a camera and in response to determining the selected image processing operation, a higher resolution core image of the rock core;
- and
- determining a permeability or a porosity of the rock core using the higher resolution core image.
11 more elements
A well system, comprising:
- a wellbore penetrating a subterranean formation at a well site;
- a well control system of the wellbore;
- and
- a core image analyzer comprising a processor and a memory, wherein the core image analyzer is configured to perform a method comprising:
- capturing a training dataset comprising a plurality of coarse core images of a plurality of rock cores and a plurality of rock core labels associated with the plurality of coarse core images;
- performing a training operation on a convolutional neural network using the training dataset and a regression algorithm to produce a trained model,
- wherein the trained model determines one or more predicted image scores in response to receiving one or more core images, and
- wherein the one or more predicted image scores are used to determine a selection among a plurality of image processing operations;
- determining a predicted image score using the trained model and a coarse core image of a rock core;
- determining a selected image processing operation among the plurality of image processing operations based on the predicted image score,
- wherein the selected image processing operation corresponds to an image acquisition of the rock core at a higher resolution than the coarse core image;
- and
- capturing, by a camera and in response to determining the selected image processing operation, a higher resolution core image of the rock core;
- and
- determining a permeability or a porosity of the rock core using the higher resolution core image.
12 more elements
Dependencies are inferred from the claim text.