This report investigates the use of machine learning packages to reduce acoustic power generated from a scramjet engine. Hypersonic vehicles travelling at high speeds inherently produce large pressure differentials, such as shock waves, which make them impractical for regulatory use, particularly over land. In this study, neural networks were trained on virtual flow simulations of different scramjet setups travelling at hypersonic speeds to predict how to maintain desirable air speed conditions while optimizing a reduction in acoustic power. The results demonstrate that computer learning approaches to engineering design problems can efficiently offer optimal solutions that lead to the quieter high-speed travel of tomorrow. In my continuing work, I will be furthering this research by finding the relationship between shock wave energy and varying distances between inlet angles, number of angles, and varying shapes of the scramjet inlet.
Ethan J. Gaul (Wed,) studied this question.