Two-stage Optimization of a Tapered Elastic Swimmer Using a Parameterized Genetic Algorithm

Fish are able to swim very efficiently at high speeds, even though most of their body is not being actuated. Part of what allows them to do this is the tapering in their bodies near their tail, which changes the size and phase of the oscillations at each point. By tailoring the shape of the tapering, we can increase the thrust and efficiency of an oscillating hydrodynamic plate. To search through the design space without resorting to full fluid-structure interaction (FSI) models, we use an evolutionary genetic algorithm to test and optimize the tapering shapes. As this model cannot directly calculate power and thrust, we use parameters that are correlated with swimming performance, including tip displacement, standing wave ratio, and displacement area. The best-performing shapes are validated using the FSI model, and gradient descent is done on these shapes to determine the optimal weight function.