AI-Driven Analysis and Characterization of Hypersonic Plasma Flows for Wind Tunnel Testing

The characterization of hypersonic flows remains a major challenge in fluid dynamics, as both experimental and numerical approaches present significant limitations. High-enthalpy experiments provide valuable but often sparse and noisy measurements, while high-fidelity CFD simulations are computationally demanding and difficult to extend over large parametric ranges.This work proposes a hybrid Artificial Intelligence framework based on Deep Neural Networks to reconstruct key flow-field quantities from limited and heterogeneous data. The model is trained on a combined dataset including high-fidelity numerical (CFD and spectral emissions) results and experimental measurements from the SPES facility at the University of Naples Federico II, enabling a connection between numerical predictions and experimental observations.Validation on unseen cases shows that the model can predict the main thermochemical quantities within a ±10% error range. These results demonstrate the potential of AI-based approaches as efficient tools to support the analysis, optimization, and planning of future hypersonic test campaigns.