Reduced Multifidelity Low-Sonic-Boom Design for Supersonic Transport by Manifold Alignment

To meet the strict economic and environmental requirements of next-generation supersonic transport, high-fidelity design is essential but computationally costly. This paper proposes an efficient low-sonic-boom design method driven by multifidelity near-field overpressure data fusion. The method employs Principal Component Analysis (PCA) for dimensionality reduction and Procrustes analysis for manifold alignment to fuse low- and high-fidelity data in a shared latent space. A multifidelity surrogate model is constructed, driven by massive low-fidelity data and corrected by limited high-fidelity data to enable accurate and fast prediction. An optimization framework is established with the Euclidean distance to the target low-boom overpressure distribution as the objective and aerodynamic performance as the constraints. Results show that the method reduces the sonic boom by 2.34 PLdB, increases the lift coefficient by 7.47%, and improves the lift-to-drag ratio by 6.96%, realizing efficient and high-precision low-sonic-boom design.