Frequency-Dependent ML for Seismic Fault Interpretation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Interpreting 3D seismic data for identifying faults is time-consuming and often biased by user expertise, with limited resolution in fault probability and the presence of false-positive faults hindering accurate identification of geologically plausible faults.
Innovation Solution
Implementing a frequency-dependent machine-learning model with aleatoric uncertainty analysis, which involves applying spectral decomposition to pre-processed training data to generate frequency-dependent training data, training multi-channel, multi-scale convolutional neural networks, and using aleatoric uncertainty analysis to filter out low-probability faults and generate high-fidelity fault probability maps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If deep learning techniques are used to automate fault predictions from seismic data, then productivity is improved, but measurement precision deteriorates due to limited resolution of fault probability and presence of false-positive faults
Solution Approach 1:
The patent segments the fault prediction process into multiple frequency-dependent analyses. By decomposing seismic data into different frequency components and applying separate machine learning models to each frequency band, the system achieves both automation and improved precision. Each frequency-specific model captures distinct geological features, reducing false positives while maintaining high productivity through systematic automated processing.
2Measurement precision
If traditional fault interpretation methods are used, then measurement precision is maintained through expert analysis, but productivity deteriorates due to time-consuming manual interpretation
Solution Approach 1:
The patent replaces manual mechanical interpretation processes with automated machine learning systems. Multiple frequency-dependent machine learning models automatically analyze seismic data, substituting human expert analysis with computational algorithms that process data faster while maintaining or improving accuracy through multi-frequency validation and uncertainty quantification.
3Measurement precision
If frequency-dependent machine learning models are applied, then measurement precision is improved through reduced false-positive faults, but device complexity increases due to multi-channel, multi-scale convolutional neural networks
Solution Approach 1:
The patent adds the frequency dimension to the fault prediction analysis by implementing multiple frequency-dependent machine learning models. Each model operates on a specific frequency band, and their results are integrated to produce a comprehensive fault probability map. This dimensional expansion improves precision by capturing frequency-specific geological features while managing complexity through modular model architecture and systematic integration.
Data Source
AI summary
Frequency-dependent machine-learning (ML) models can be used to interpret seismic data. A system can apply spectral decomposition to pre-processed training data to generate frequency-dependent training data of two or more frequencies. The system can train two or more ML models using the frequency-dependent training data. Subsequent to training the two or more ML models, the system can apply the two or more ML models to seismic data to generate two or more subterranean feature probability maps. The system can perform an analysis of aleatoric uncertainty on the two or more subterranean feature probability maps to create an uncertainty map for aleatoric uncertainty. Additionally, the system can generate a filtered subterranean feature probability map based on the uncertainty map for aleatoric uncertainty.