Carbon Conversion

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)
  • Optimize compatible materials that maximize electrochemical CO2 conversion
  • Simulate nano-bubbles to facilitate biological update
  • Search literature on cost effective algal strains for high-volume CO2 conversion
  • Determine the optimal strain for CO2 uptake in a specific project
  • Identify pre-treatment strategies that enhance photosynthetic efficiency
  • Identify optimum locations and algal processes to maximize benefits
  • Obtain information about site optimization
  • Build robust data sets to better characterize waste materials for mineralization
  • Predict decomposition/deactivation behaviors in biomass
  • Ensure proactive maintenance and integrity of infrastructure
  • Monitor long-term durability of carbon removal
  • Structured numeric data, text, image, maps, sequences, time-series data
  • Discover New Catalysts or Improve Current Catalysts
  • Novel algal CO2 delivery systems.
  • Develop/Demonstrate integrated CO2 capture & advanced algal concepts.
  • Geospatial Optimization
  • Optimize process design for CDR & cost
Approaches
Use machine learning approaches without known labels/expectations, usually for
  • Clustering
  • Dimension reduction
  • Determine optimum configurations/reaction pathways for target end products
  • Identify high-intensity, non-traditional products that can be derived from CO2
  • Identify pre-treatment strategies that enhance photosynthetic efficiency
  • Optimize resource efficiency for mineralization
  • Build robust data sets to better characterize waste materials for mineralization
  • Structured or unstructured data
  • Discover New Catalysts or Improve Current Catalysts
  • Produce non-traditional products from CO2 conversion
  • Characterize waste products
  • Develop/Demonstrate integrated CO2 capture & advanced algal concepts.
Approaches
Use complex neural network architecture for
  • Image and spatial prediction
  • Time series modeling/prediction
  • Natural language processing (NLP)
Can also be used for Transfer Learning or Reinforcement Learning
  • Develop NN to predict catalyst performance & durability
  • Design/Predict novel catalyst/simulate catalysis processes and products from atomic level
  • Predict decomposition/deactivation behaviors in biomass
  • Collect data to formulate required probability distributions for Monte Carlo type simulation of economic (TEA) and emission (LCA) performance of various conversion technologies
  • Structured numeric data, text, image, maps, sequences, time-series data
  • Discover New Catalysts or Improve Current Catalysts
  • Develop/Demonstrate integrated CO2 capture & advanced algal concepts.
Approaches
Combine more than one machine learning model to handle the tasks (tree-based, such as Random Forest, XGBoost, or deep learning, such as Committee Machine)
  • Predict or design optimal reactors and processes for CO2 conversion pathways (w/CFD)
  • Optimize the genes of algae strains or plants to achieve high CO2 retention
  • For mid-TRL, improve “smart control” of nutrient release and cultivation
  • Optimize resource efficiency for mineralization
  • Build robust data sets to better characterize waste materials for mineralization
  • Optimize mineralization process based on incoming materials and desired product properties
  • Explore CO2 use in advanced materials and alloys
  • Structured numeric data, text, image, maps, sequences, time-series data
  • Produce non-traditional products from CO2 conversion
  • Novel algal CO2 delivery systems.
  • Develop/Demonstrate integrated CO2 capture & advanced algal concepts.
  • Characterize waste products
  • Develop & Optimize processes that integrate CO2 capture with mineral carbonization.
  • LCA & TEA (Techno-economic assessments)
  • Apply CO2 in Adv. Manufacturing
Approaches

Integrate scientific equations and knowledge into machine learning algorithms (usually using Deep Learning), such as Physics-Informed Neural Networks

  • Predict or design optimal reactors and processes for CO2 conversion pathways (w/CFD)
  • Optimize the genes of algae strains or plants to achieve high CO2 retention
  • Optimize compatible materials that maximize electrochemical CO2 conversion.
  • Simulate nano-bubbles to facilitate biological update
  • Predict decomposition/deactivation behaviors in biomass
  • Structured numeric data, text, image, maps, sequences, time-series data
  • Scientific equations, field equations, partial differential equations
  • Develop & Optimize processes that integrate CO2 capture with mineral carbonization.
  • Novel algal CO2 delivery systems.
  • Develop/Demonstrate integrated CO2 capture & advanced algal concepts.