A Method for Flow Field Reconstruction based on Fourier Neural Operator Network.

Flow field reconstruction is a key task of aircraft sensing the external environment. Numerical methods like CFD are time-consuming and computationally expensive, so surrogate model-based methods are adopted in the task. However, most kinds of surrogate model only learn the mapping from the input space to the output space. Fourier neural operator (FNO) is a model structure that can learn from the functional input space to output functional space. This paper develops the FNO method to reconstruct the flow field around a wing using pressure sensor measurements on the wing surface. We carried out experiments to validate the developed method, and the result shows our method can achieve high accuracy and noise robustness.