Sunwoong Yang

▼ Key components of Physics-Guided AI

1️⃣ Data Efficiency and Practicality 

• Trustworthiness requires special data handling strategies (e.g., multi-fidelity modeling) for reliable performance despite the limitations of sparse/imperfect practical datasets

2️⃣ Physical Consistency

• Trustworthiness requires that models respect established physical laws, ensuring predictions are realistic and preventing non-physical outcomes  

3️⃣ Robustness in Spatio-Temporal Prediction

• For dynamic systems, trustworthiness requires that predictions remain stable and accurate over long time horizons, effectively mitigating error accumulation 

4️⃣ Reliable Uncertainty Quantification (UQ)

• Trustworthiness depends on a model accurately quantifying its own uncertainty, enabling engineers to make informed, risk-aware decisions 

▼ Specialization of Physics-Guided AI for Aerodynamic Simulation & Design 

My research systematically integrates above four key components into a unified end-to-end framework,
providing a roadmap as below for developing physics-guided AI in aerodynamic simulation & design

• Title: Development of AI-based Flow Prediction Framework Considering Versatility and User-Friendliness in Digital Twins
  Sponsor: National Research Foundation of Korea (Sejong Fellowship)
  Period: 2024.05.01 ~ 2029.04.30 (5 years, 500,000,000 KRW)
  Role: PI

• Title: Development of AI algorithms for preform design for blow molding
  Sponsor: Korea Institute of Industrial Technology (KITECH)
  Period: 2024.05.08 ~ 2024.08.31 (4 months, 20,000,000 KRW)
  Role: PI

Hope you enjoy exploring!