Rapid flow field prediction in patterned baleen membranes of balaenid whales during filter feeding by deep learning.

This article focuses on the development and validation of a 3-input, 9-output UNet deep learning framework, named UNet-BaleenCFD, for rapid flow field prediction in patterned baleen membranes of balaenid whales during filter feeding. Using computational fluid dynamics (CFD) simulations combined with linear interpolation, the study generates datasets to train the model, which predicts velocity and pressure fields with accuracy comparable to CFD but at speeds three orders of magnitude faster. The research demonstrates that different patterned baleen membranes influence flow and pressure distributions uniquely, and the UNet-BaleenCFD model shows good generalization ability beyond the training data. This approach offers theoretical guidance for designing biomimetic filtration membranes inspired by balaenid whales, with potential applications in water purification, sewage treatment, and oil filtration.