Back to results
US 12,742,374 B2Granted
Machine learning approach for generating subsurface representations in layer-space
AssigneeChevron U.S.A. Inc.
InventorsLewis Li, Tao Sun
Life of this patent
Estimating the term…
Abstract
Subsurface representations that define subsurface configurations in layer space are used to train a machine learning model. The trained machine learning model is used to generate synthetic subsurface representations in the layer space. The synthetic subsurface representations are generated to match one or more conditioning characteristics. Conditioning of the trained machine learning model is performed in latent space.
Claims20
A system for generating synthetic subsurface representations, the system comprising:
- one or more physical processors configured by machine-readable instructions to:
- obtain subsurface representation information, the subsurface representation information defining a set of subsurface representations, the set of subsurface representations defining subsurface configurations in layer space such that the subsurface configurations are defined as a function of layers, wherein vertical dimension in the layer space is layers rather than physical distance, wherein a given subsurface representation for a given subsurface region with multiple physical layers of different thicknesses represent the multiple physical layers of different thicknesses with uniform layers in the layer space;
- train a machine learning model using the set of subsurface representations within the layer space, wherein the trained machine learning model generates the synthetic subsurface representations in the layer space;
- and
- store the trained machine learning model in a storage medium.
2 more elements
A method for generating synthetic subsurface representations, the method comprising:
- obtaining subsurface representation information, the subsurface representation information defining a set of subsurface representations, the set of subsurface representations defining subsurface configurations in layer space such that the subsurface configurations are defined as a function of layers, wherein vertical dimension in the layer space is layers rather than physical distance, wherein a given subsurface representation for a given subsurface region with multiple physical layers of different thicknesses represent the multiple physical layers of different thicknesses with uniform layers in the layer space;
- training a machine learning model using the set of subsurface representations within the layer space, wherein the trained machine learning model generates the synthetic subsurface representations in the layer space;
- and
- storing the trained machine learning model in a storage medium.
1 more element
Dependencies are inferred from the claim text.