Directing Shallow-Water Waves Using Fixed Varying Bathymetry Designed by Recurrent Neural Networks

Directing shallow-water waves and their energy is highly desired in many ocean engineering applications. Coastal infrastructures can be protected by reflecting to deep water. Wave harvesting efficiency improved focusing on wave converters. Changing water depth effectively affect celerity therefore the propagation of waves. However, determining spatially varying bathymetry that direct a designed location not trivial. In this paper, we propose novel machine learning method design optimize for directing waves, which assumed fixed time without considering morphodynamics. Shallow-water theory was applied establish mapping between mechanics recurrent neural networks (RNNs). Two wave-equivalent RNNs were developed model over bathymetry. The resulting trained focusing. We demonstrate optimized reflect refract various locations. also foresee potential new tools similarly based mathematical equivalence networks.