Automatic model selection through machine learning
A method can include receiving data for a geologic region; based at least in part on the data, selecting a model from a plurality of models using a trained machine learning model, and inverting the data using the selected model to determine parameters of the selected model.
1 . A method comprising:
receiving real-time data for a geologic region, wherein the real-time data is acquired via a tool disposed in a borehole in the geologic region, and wherein the geologic region comprises a plurality of subregions, each subregion corresponding to different structural features of the geologic region;
determining one or more first geometric attributes for a first subregion of the plurality of subregions based on the real-time data;
selecting, based at least in part on the one or more first geometric attributes, a first model from a plurality of models using a trained machine learning model, wherein the plurality of models comprises at least two forward inversion models having distinct dimensionalities, and wherein the first model corresponds to a first forward inversion model of the at least two forward inversion models and is selected based on a dimensionality corresponding to the one or more first geometric attributes;
inverting, for the first subregion, the real-time data using the first model to determine one or more first subsurface parameters associated with the first subregion;
determining that the one or more first subsurface parameters do not converge relative to the real-time data;
selecting, based at least in part on the one or more first geometric attributes, a second model from the plurality of models using the trained machine learning model in response to the one or more first subsurface parameters not converging, wherein the second model corresponds to a second forward inversion model having a higher dimensional complexity relative to the first forward inversion model;
inverting, for the first subregion, the real-time data using the second model to determine one or more second subsurface parameters associated with the first subregion;
determining that the one or more second subsurface parameters converge relative to the real-time data;
consolidating the one or more second subsurface parameters from the first subregion to generate a model for the geologic region in response to the one or more second subsurface parameters converging; and
controlling drilling equipment within the borehole based on the model.
2 . The method of claim 1 , wherein the trained machine learning model comprises a support vector machine model.
3 . The method of claim 1 , wherein the trained machine learning model comprises a k-nearest neighbors model.
4 . The method of claim 1 , wherein the trained machine learning model comprises a classifier model.
5 . The method of claim 1 , wherein the trained machine learning model comprises a neural network model.
6 . The method of claim 1 , wherein the real-time data comprises logging while drilling data.
7 . The method of claim 1 , wherein the real-time data comprises resistivity data.
8 . The method of claim 1 , comprising biasing the selection of the first model towards a lower model dimensionality relative to a respective dimensionality of the second model.
9 . The method of claim 8 , wherein the lower model dimensionality corresponds to a higher computational efficiency.
10 . The method of claim 1 , comprising simulating physical phenomena using the model.
11 . The method of claim 1 , comprising rendering a graphical user interface to a display that comprises a graphical control for selecting the first model, the second model, or both.
12 . The method of claim 1 , comprising training the machine learning model.
13 . The method of claim 12 , wherein the training comprises supervised learning.
14 . The method of claim 1 , wherein the one or more first subsurface parameters comprise one or more petrophysics parameters.
15 . The method of claim 1 , wherein the real-time data corresponds to a window size.
16 . The method of claim 15 , wherein the window size is adjustable.
17 . A system comprising:
a processor;
memory operatively coupled to the processor; and
processor-executable instructions stored in the memory to instruct the system to:
receive real-time data for a geologic region, wherein the real-time data is acquired via a tool disposed in a borehole in the geologic region, and wherein the geologic region comprises a plurality of subregions, each subregion corresponding to different structural features of the geologic region;
determine one or more first geometric attributes for a first subregion of the plurality of subregions based on the real-time data;
select, based at least in part on the one or more first geometric attributes, a first model from a plurality of models using a trained machine learning model, wherein the plurality of models comprises at least two forward inversion models having distinct dimensionalities, and wherein the first model corresponds to a first forward inversion model of the at least two forward inversion models and is selected based on a dimensionality corresponding to the one or more first geometric attributes;
invert, for the first subregion, the real-time data using the first model to determine one or more first subsurface parameters associated with the first subregion;
determine that the one or more first subsurface parameters do not converge relative to the real-time data;
select, based at least in part on the one or more first geometric attributes, a second model from the plurality of models using the trained machine learning model in response to the one or more first subsurface parameters not converging, wherein the second model corresponds to a second forward inversion model having a higher dimensional complexity relative to the first forward inversion model;
invert, for the first subregion, the real-time data using the second model to determine one or more second subsurface parameters associated with the first subregion;
determine that the one or more second subsurface parameters converge relative to the real-time data;
consolidate the one or more second subsurface parameters from the first subregion to generate a model for the geologic region in response to the one or more second subsurface parameters converging; and
control drilling equipment within the borehole based on the model.
18 . The system of claim 17 , wherein the processor-executable instructions instruct the system to bias the selection of the first model towards a lower model dimensionality relative to the second model.
19 . One or more non-transitory computer-readable storage media comprising computer-executable instructions executable to instruct a computing system to:
receive real-time data for a geologic region, wherein the real-time data is acquired via a tool disposed in a borehole in the geologic region, and wherein the geologic region comprises a plurality of subregions, each subregion corresponding to different structural features of the geologic region;
determine one or more first geometric attributes for a first subregion of the plurality of subregions based on the real-time data;
select, based at least in part on the one or more first geometric attributes, a first model from a plurality of models using a trained machine learning model, wherein the plurality of models comprises at least two forward inversion models having distinct dimensionalities, and wherein the first model corresponds to a first forward inversion model of the at least two forward inversion models and is selected based on a dimensionality corresponding to the one or more first geometric attributes;
invert, for the first subregion, the real-time data using the first model to determine one or more first subsurface parameters associated with the first subregion;
determine that the one or more first subsurface parameters do not converge relative to the real-time data;
select, based at least in part on the one or more first geometric attributes, a second model from the plurality of models using the trained machine learning model in response to the one or more first subsurface parameters not converging, wherein the second model corresponds to a second forward inversion model having a higher dimensional complexity relative to the first forward inversion model;
invert, for the first subregion, the real-time data using the second model to determine one or more second subsurface parameters associated with the first subregion;
determine that the one or more second subsurface parameters converge relative to the real-time data;
consolidate the one or more second subsurface parameters from the first subregion to generate a model for the geologic region in response to the one or more second subsurface parameters converging; and
control drilling equipment within the borehole based on the model.
20 . The one or more non-transitory computer-readable storage media of claim 19 , wherein the computer-executable instructions instruct the computing system to bias the selection of the first model towards a lower model dimensionality relative to the second model.