Data Driven Aerodynamics

Deep Learning Applications to Airfoil Performance Prediction and Inverse Design, Deep ConvNets, cGANs

A Convolutional Neural Network Approach for Training Predictors for Airfoil Performance

  • Medium-fidelity predictors were trained for predicting airfoil performance using deep convolutional neural networks. The framework was implemented in Tensorflow.
  • Published a conference paper: AIAA Aviation’17.
The Application of Deep Convolutional Neural Networks to Airfoil Performance Prediction

A Deep Learning Approach to an Airfoil Inverse Desing Problem

  • Airfoil inverse problem was reformulated and solved using a framework constructed with deep convolutional neural networks.
  • Published a conference paper: AIAA Aviation’18.
The Application of Deep Convolutional Neural Networks to Airfoil Inverse Design

Conditional Generative Adversarial Network Framework for Airfoil Inverse Design

  • A game theoretic framework was investigated by training conditional generative adversarial networks (CGANs) to generate airfoil shapes given the desired airfoil performance criteria.
  • Published a conference paper: AIAA Aviation’20.
  • Funding Proposal Experience
    -ARPA-E: DIFFERENTIATE (Design Intelligence Fostering Formidable Energy Reduction and Enabling Novel Totally Impactful Advanced Technology Enhancements) Program.
Airfoil Shape Generation by the Generator in GAN Framework Shown at Different Epochs

References