FLRNet, a VAE-based deep network with Fourier features and perceptual loss, reconstructs cylinder-wake flow fields from 8-32 sensors more accurately than POD and MLP baselines in the reported tests.
Deep-learning assisted reduced order model for high-dimensional flow prediction from sparse data
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
physics.flu-dyn 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
FLRNet: A Deep Learning Method for Regressive Reconstruction of Flow Field From Limited Sensor Measurements
FLRNet, a VAE-based deep network with Fourier features and perceptual loss, reconstructs cylinder-wake flow fields from 8-32 sensors more accurately than POD and MLP baselines in the reported tests.