Recent publication on machine learning in turbulence
Check out our recent article "On the benefits and limitations of Echo State Networks for turbulent flow prediction". Together with Jörg Schumacher's group, we have attempted to predict the unsteady von Kármán vortex street both statistically and deterministically. We have evaluated up to which extent the Echo State Network is capable of predicting the flow and which hyperparameters should be chosen for each approach. This is the first step in our attempt to predict or model turbulent flows in a broader context.
Check out the full paper (open access) here:
M. Sharifi Gazjahani, F. Heyder, J. Schumacher, C. Cierpka (2022) On the benefits and limitations of Echo State Networks for turbulent flow prediction, Measurement Science and Technology 34, 014002, DOI: 10.1088/1361-6501/ac93a4, open access