Super-resolution reconstruction of high-fidelity rotating detonation flow fields based on cycle-consistent generative adversarial networks

This Letter presents a Cycle-Consistent Generative Adversarial Network (CycleGAN)-based framework with additional content and gradient reconstruction losses to achieve 8× super-resolution reconstruction of rotating detonation engine (RDE) flow fields. When trained and validated on OpenFOAM-simulated hydrogen-air RDE datasets, the method recovers fine-scale features with high fidelity. Qualitative and quantitative evaluations demonstrate its efficacy in reconstructing critical detonation wave structures.