Adjoint-based machine learning for active flow control

Xuemin Liu, Jonathan F. MacArt

We develop neural-network active flow controllers using a deep learning PDE augmentation method (DPM). The sensitivities for optimization are computed using adjoints of the governi…