Multi-Fidelity Surrogate-Based Trajectory-Aware Shape Optimization of Hypersonic Vehicles

This work presents a trajectory-aware optimization framework for hypersonic glide vehicles that couples adjoint-based sensitivity analysis with multi-fidelity Gaussian processes. To address the computational cost of hypersonic flow simulations, a continuous-time adjoint formulation is used to identify specific trajectory segments where aerodynamic perturbations most strongly influence vehicle range. This sensitivity profile drives a targeted sampling strategy that allocates high-fidelity CFD evaluations only to the most influential flight conditions, fusing them with low-fidelity data to construct accurate aerodynamic surrogates. The framework is applied to optimize the nose radius of a glide vehicle via Bayesian optimization, balancing range performance against peak heat flux constraints. Numerical results demonstrate that the proposed method effectively concentrates computational resources in critical flight phases, yielding accurate load predictions and a thermally compliant, range-optimized design with a limited budget of high-fidelity evaluations.