Design exploration of hypersonic air-breathing vehicle including airframe and air inlet using deep-learning flowfield prediction

Abstract Hypersonic airbreathing vehicles are promising candidates for high-speed point-to-point transportation and first stage of space transportation system. Airbreathing engines are often mounted under the vehicle to utilize the compression of shock waves formed by the vehicle shape. However, the design of airbreathing engines and vehicles’ airframe are often conducted in a separate manner. This does not always lead to a global optimization of hypersonic vehicle design considering aero-propulsive balance and stability. The present study develops a fast approach for full-vehicle aero-propulsive performance evaluation to realize multi-objective optimization of hypersonic vehicle considering aero-propulsive interaction. To provide a global reconstruction of the coupled flowfield behavior across the design space, a deep-learning model is employed to predict scramjet flowfields due to its capability to capture strongly nonlinear features with high computational efficiency for high-dimensional data such as flowfields, while a local surface inclination method is used to evaluate the aerodynamic performance of the airframe. The proposed framework is applied to a representative multi-objective optimization problem, demonstrating its capability for simultaneous airframe–propulsion design exploration.