Data-driven global optimization and mechanism analysis of flapping motion for a hybrid underwater robot

This study focuses on the flapping propulsion performance of a new type of hybrid-driven underwater robot. For this specific configuration of the robot, the optimal motion parameters cannot be directly obtained from biomimetic approaches. Therefore, we systematically analyzed the influence of pectoral fin motion parameters (frequency, phase difference, and amplitude) on thrust and lift characteristics using computational fluid dynamics (CFD), revealing the independent effects of each parameter. Based on the CFD simulation data, a machine learning model based on the Light Gradient Boosting Machine was developed to establish an accurate mapping between the motion parameters and hydrodynamic performance. The model demonstrates high accuracy in predicting thrust and lift coefficients, which satisfies the requirements of global optimization. The global optimal set of motion parameters for this specific robot configuration was obtained through optimization algorithms. Furthermore, by visualizing the vortex structures and pressure cloud maps of the flow field under optimal conditions, the origin of its superior hydrodynamic performance was interpreted from the perspective of flow mechanisms. The results demonstrate that collaborative parameter optimization can significantly improve propulsion efficiency and motion stability. This study provides an important theoretical basis and data support for motion control and performance optimization of this type of hybrid-driven underwater robot.