Research on optimal stiffness distribution of homocercal fish tail based on surrogate modeling.

The focus of this work is to investigate the influence of stiffness distribution in the fish tail on swimming performance and to determine the optimal stiffness distribution. Targeting fish employing the body and/or caudal fin (BCF) swimming mode, we constructed an fluid-structure interaction (FSI) simulation model based on the characteristics of BCF locomotion. Using this FSI model, we systematically examined multiple typical stiffness distributions along the inter-ray and ray-aligned directions, summarizing the underlying patterns in these two directions. Subsequently, we expanded the dataset obtained from the FSI simulations. Based on the expanded dataset, we developed a surrogate model using support vector regression (SVR) enhanced by the young's double-slit experiment optimization algorithm (YDSE). An improved particle swarm optimization algorithm was then applied to this surrogate model to identify the stiffness distributions corresponding to maximum thrust and highest efficiency, respectively. Compared to the original dataset, the optimized solutions obtained through YDSE-SVR iteration increased thrust by 4.94% and efficiency by 6.86%. Finally, we analyzed the mechanisms behind the differences in thrust and efficiency using pressure contours and streamline diagrams. The derived patterns regarding the influence of fish tail stiffness distribution on swimming performance can provide insights for robotic fish design.