Check out our recent open-access publication in Machine Learning: Science and Technology "On the prediction of the turbulent flow behind cylinder arrays via Echo State Networks". We show that Echo State Networks (ESNs) can model the turbulent flows and the complex interactions of vortex streets behind cylinder arrays, even over extended periods in a closed-loop scenario.  Given the quality of predictions by the ESN in the current study for such chaotic dynamics, ESNs and, more generally, reservoir computing can receive more attention from the community in comparison to much more complex algorithms that might not necessarily result in better outcomes.

 

Reference: Mohammad Sharifi Ghazijahani, Christian Cierpka; On the prediction of the turbulent flow behind cylinder arrays via Echo State Networks. Machine Learning: Science and Technology 2024; 5, 035005: https://iopscience.iop.org/article/10.1088/2632-2153/ad5414