Geophysical deep learning

Patent metadata

Field	Value
Publication number	US-11313994-B2
Application number	US-201816484879-A
Country	US
Kind code	B2
Filing date	Feb 9, 2018
Priority date	Feb 9, 2017
Publication date	Apr 26, 2022
Grant date	Apr 26, 2022

Abstract

A method can include selecting a type of geophysical data; selecting a type of algorithm; generating synthetic geophysical data based at least in part on the algorithm; training a deep learning framework based at least in part on the synthetic geophysical data to generate a trained deep learning framework; receiving acquired geophysical data for a geologic environment; implementing the trained deep learning framework to generate interpretation results for the acquired geophysical data; and outputting the interpretation results.

First claim

What is claimed is:

1. A method comprising:
selecting a type of geophysical data;
selecting a type of algorithm;
generating synthetic geophysical data of the selected type of geophysical data based at least in part on the algorithm;
training a deep learning classifier framework based at least in part on the synthetic geophysical data to generate a trained deep learning classifier framework;
receiving acquired geophysical data for a geologic environment;
implementing the trained deep learning classifier framework to generate interpretation results for the acquired geophysical data; and
outputting the interpretation results, wherein the interpretation results characterize structural features indicative of hydrocarbons in the geologic environment.

2. The method of claim 1 wherein the type of geophysical data comprises seismic trace data.

3. The method of claim 1 wherein the type of geophysical data comprises log data acquired via a downhole tool.

4. The method of claim 1 wherein the type of geophysical data comprises surface controlled electromagnetic data.

5. The method of claim 1 wherein the type of algorithm comprises a model modification algorithm that randomly modifies at least a portion of a model of the geologic environment within a predefined set of parameters for generating one or more portions of the synthetic geophysical data.

6. The method of claim 1 wherein the type of algorithm comprises a filtering algorithm.

7. The method of claim 6 wherein the filtering algorithm comprises a frequency-domain filtering algorithm or a time-domain filtering algorithm and wherein the type of geophysical data comprises seismic trace data, wherein the frequency-domain filtering algorithm comprises performing a Fourier transform on a seismic trace to output an amplitude spectrum, multiplying the amplitude spectrum by an amplitude spectrum of a filter operator to generate a result and performing an inverse Fourier transform on the result to output a filtered seismic trace as a portion of the synthetic geophysical data, and wherein the time-domain filtering algorithm comprises performing an inverse Fourier transform on an amplitude spectrum to generate a filter operator and convolving the filter operator with a seismic trace to output a filtered seismic trace as a portion of the synthetic geophysical data.

8. The method of claim 6 wherein the filtering algorithm comprises a spatial filtering algorithm.

9. The method of claim 1 wherein the type of algorithm comprises a noise generation algorithm.

10. The method of claim 9 wherein the type of geophysical data comprise seismic trace data, wherein the noise generation algorithm comprises a coherent noise generation algorithm that generates synthetic coherent noise or that extracts coherent noise from real seismic trace data acquired via a seismic survey.

11. The method of claim 1 wherein the type of geophysical data comprise seismic trace data, wherein the type of algorithm comprises an acquisition geometry variation algorithm that perturbs an acquisition geometry of a seismic survey to generate one or more perturbed acquisition geometries, wherein the generating synthetic geophysical data comprises simulating seismic trace data with the one or more perturbed acquisition geometries.

12. The method of claim 1 comprising combining the synthetic geophysical data with at least a portion of the acquired geophysical data and training the deep learning framework based at least in part on the combined geophysical data.

13. The method of claim 1 wherein generating synthetic geophysical data based at least in part on the algorithm generates a plurality of sets of synthetic geophysical data wherein each of the sets differs with respect to an acquisition parameter value.

14. The method of claim 1 wherein generating synthetic geophysical data based at least in part on the algorithm generates a plurality of sets of synthetic geophysical data wherein each of the sets differs with respect to a processing parameter value.

15. The method of claim 1 wherein generating synthetic geophysical data based at least in part on the algorithm generates a plurality of sets of synthetic geophysical data wherein each of the sets differs with respect to a geology parameter value.

16. A system comprising:
a processor;
memory operatively coupled to the processor; and
processor-executable instructions stored in the memory to instruct the system to:
select a type of geophysical data;
select a type of algorithm;
generate synthetic geophysical data of the selected type of geophysical data based at least in part on the algorithm;
train a deep learning classifier framework based at least in part on the synthetic geophysical data to generate a trained deep learning classifier framework;
receive acquired geophysical data for a geologic environment;
implement the trained deep learning classifier framework to generate interpretation results for the acquired geophysical data; and
output the interpretation results, wherein the interpretation results characterize structural features indicative of hydrocarbons in the geologic environment.

17. The system of claim 16 wherein the type of geophysical data comprises seismic trace data.

18. The system of claim 16 wherein the type of geophysical data comprises log data.

19. The system of claim 16 comprising an interface that outputs at least one control signal based at least in part on the interpretation results.

20. One or more non-transitory computer-readable storage media comprising processor-executable instructions to instruct a computing system to:
select a type of geophysical data;
select a type of algorithm;
generate synthetic geophysical data of the selected type of geophysical data based at least in part on the algorithm;
train a deep learning classifier framework based at least in part on the synthetic geophysical data to generate a trained deep learning classifier framework;
receive acquired geophysical data for a geologic environment;
implement the trained deep learning classifier framework to generate interpretation results for the acquired geophysical data; and
output the interpretation results, wherein the interpretation results characterize structural features indicative of hydrocarbons in the geologic environment.

Assignees

• Schlumberger Technology Corp

Inventors

• Salman Nader
• Aarre Victor
• Borgos Hilde Grude
• Nickel Michael Hermann

Classifications

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G01V2210/614

  Synthetically generated data · CPC title

• G01V1/282 Primary

  Application of seismic models, synthetic seismograms · CPC title

• G06N3/04

  Architecture, e.g. interconnection topology · CPC title