基于舵机驱动的绳驱仿生鱼尾动力学建模与实验验证

Dynamic Modeling and Experiment of Cable-driven Bionic Fishtail Based on Servo Drive

摘要:

绳驱仿生鱼因其结构简单、柔顺性高等优势,已成为海洋精细探测平台的重要研究方向。 该类仿生鱼通常采用位于鱼尾首部的舵机转动来驱动鱼尾摆动,从而实现推进。 然而,这种单点驱动方式难以直接适用于经典的、基于独立节点控制的鱼体波函数。 针对该问题,本文提出一种改进的鱼体波构造函数,该函数通过直接建立鱼尾摆动角度与舵机旋转角度之间的几何关系进行建模,进而构建了仿生鱼尾的拉格朗日动力学模型。 该模型可根据已知的舵机旋转角度和频率,预测仿生鱼尾的推进性能。 通过 ANSYS Fluent 仿真获取水动力参数,并结合水池试验验证了模型的准确性。 结果表明,理论预测与试验测量结果基本吻合,验证了模型的有效性。 该模型为绳驱仿生鱼的推进性能预测与控制优化提供了理论依据。

Abstract:

Cable-driven bionic fish have attracted increasing attention as promising underwater platforms for fine-scale marine exploration owing to their lightweight structure, high compliance, and low hydrodynamic disturbance. In many prototypes, a single servo motor mounted at the anterior tail joint pulls antagonistic tendons to generate a characteristic C-shaped tail beat, which differs significantly from the classical S-shaped body wave assumed in most theoretical studies. As a result, existing body wave models cannot directly describe the kinematics and propulsion performance of such cable-driven configurations. To address this gap, a geometric mapping between the servo rotation and the joint deflection of a multi-segment tail was developed, and on this basis a modified Lighthill-type body wave function whose amplitude envelope was explicitly parameterized by the servo angle and frequency was proposed. A discrete Lagrangian dynamics model of the cable-driven tail was then formulated by introducing hydrodynamic drag, lift, and added-mass forces obtained from CFD simulations in ANSYS Fluent with overset moving meshes. The model was used to predict the forward swimming speed of a six- segment robotic tail under different combinations of tail-beat frequency (1 ~ 3 Hz )and maximum oscillation amplitude (25° ~ 30°). Water-tank experiments on a fully assembled robotic fish prototype, driven by an Arduino-based sinusoidal servo control, were conducted to validate the theoretical predictions. For low-frequency cases, the predicted propulsion speeds showed good agreement with both CFD and experimental results, with only small deviations observed, whereas larger discrepancies at higher frequencies were traced to strong unsteady flow, vortex shedding, and phase lag between lift and drag coefficients that were not fully captured by the quasi-steady assumption. The proposed modeling framework therefore provided a practical tool for performance prediction, parameter optimization, and control design of C-type cable-driven robotic fish tails used in underwater exploration.