Machine Learning for Hypersonic Research
Abstract
Artificial Intelligence (AI) has shown promising results in many areas of science and engineering. However, it is vastly unexplored in the field of hypersonic research. The proposed research aims to advance computational techniques in Machine Learning (ML) (the core of AI) in the framework for hypersonic airframes. We plan to combine computational and experimental data from aerodynamic, acoustic and/or structural dynamics research using ML to discover new physics-driven patterns and phenomena occurring in the high-speed flying regimes around a hypersonic airframe. We will examine the accuracy of different ML models and provide USAF with recommendations for best practices aiming to reduce engineering uncertainty during the design and development phases. In particular, we expect that the new ML models will improve our understanding of the acoustic fatigue processes occurring on a hypersonic airframe. The developed models will enable USAF and their partners to increase the impact of their experimental and computational data, as they could predict new data points that are neither experimentally measured nor numerically obtained. In addition, the ML models will allow the data interpolation obtained from computations or experiments to project results for a broad range of parameters. This research will provide USAF with new knowledge on the potential use of ML (and AI) for hypersonic research and development. We will achieve the above through close collaboration with Air Force Research Laboratory at Wright Patterson and their domestic and international partners.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 20, 2023
- Source ID
- FA86552217026
Entities
People
- Dimitris Drikakis
Organizations
- Air Force Office of Scientific Research
- United States Air Force