Methane Mitigation

AI/ML Approaches for Research and Development Application

Approaches
Use machine learning approaches based on known labels/expectations, usually for
  • Classification (e.g., Logistic, Support Vector Machine)
  • Regression (e.g., Linear, Kernel Ridge)
  • Geophysical data inversion
  • Geochemically informed leak detection
  • Legacy well log data analysis for 3D basin model development
  • Sand production detection
  • Structured numeric data, text, image, maps, sequences, time-series data
  • Methane detection and leak location
  • Predictive and prescriptive methane operation and management
  • Predictive regulation and decision support
  • Real-time prediction and monitoring
  • Operation optimization
Approaches
Use machine learning approaches without known labels/expectations, usually for
  • Clustering
  • Dimension reduction
  • Multi-level fracture network imaging
  • Automate data discovery and integration based on geologic core properties and other features
  • Identify similarities and charateristics within the data to group data into clusters
  • Reduce the dimensions and complexities of the data
  • Structured or unstructured data
  • Anomaly detection for detecting unusual emission fluctuations and disaster prevention
  • Data reduction to identify key factors that contribute to methane emissions, thereby informing the development of targeted mitigation strategies
  • Data integration and discovery
Approaches
Use complex neural network architecture for
  • Image and spatial classification
  • Time series regression
  • Natural language processing
  • Accelerate data processing of large sensor dataset
  • Predict subsurface methane pressure and concentration
  • Structured numeric data, text, image, maps, sequences, time-series data
  • Analyze images or video footage to identify sources of methane emissions
  • Forecast future methane emissions based on historical data and environmental conditions
  • Analyze methane emissions data collected over time to identify patterns or anomalies
  • Pipeline failure prediction and prevention
  • Spatio-temporal analysis on natural gas supply chain to identify and reduce emission hotspots
Approaches
Combine more than one machine learning models to handle the tasks
  • Random Forest
  • XGBoost
  • Committee Machine
  • Predict infrastructure life and risks
  • Predict subsurface methane pressure and concentration
  • Structured numeric data, text, image, maps, sequences, time-series data
  • Combine multiple detection and forecasting models to provide more accurate and reliable predictions of leaks and future emissions, respectively
  • Predictive and prescriptive operation optimization
  • Predictive decision support
Approaches
Integrate scientific equations and knowledge into machine learning algorithms (usually using Deep Learning), such as Physics-Informed Machine Learning
  • Predict geochemical or geophysical data problem
  • Forecast subsurface methane pressure and concentration
  • Solve partial differential equations as generalized solver
  • Structured numeric data, text, image, maps, sequences, time-series data
  • Scientific equations, field equations, partial differential equations
  • Predictive and prescriptive operation optimization
  • Multi-scale and resolution prediction
  • Predictive decision support
Large Language Models for UNSTRUCTURED DATA MINING Large Language Models for GENERATIVE DISCOVERY Transfer Learning Science-based Learning Edge Computing Advance Sensor Reinforcement Learning Operator Learning
Characterize Materials
Develop Structural/Functional Materials
Enable Smart Components w/embedded sensors
Optimize regional Planning and Operations
Modeling/Simulation Bridging Scales
Accelerate Flow Sheet Initialization
Optimize Turbine Operations
Assess/Control H2 Combustion Characteristics