毕业生知识库

随着科学技术的进步与发展，研究学者在仿生机器鱼系统的机械结构设计与运动性能优化等领域取得了诸多成果，使其逐渐成为探索水下环境的重要载体。然而，与自然界的鱼类相比，仿生机器鱼在游动性能及运动效率等方面仍存在着较大的差距，需要更为深入的研究与探索。

本文围绕仿生机器鱼的系统设计与建模、游动性能优化、快速转向控制策略设计以及路径规划与跟踪控制等问题展开研究。主要研究工作内容总结如下：

一、针对现有机器鱼系统无法兼顾高游动速度与低功耗的问题，通过结合柔性材料与多关节结构设计，提出了一种多柔性关节仿生机器鱼设计方案，并研制开发了原型样机。针对多柔性关节仿生机器鱼的平面运动，基于伪刚体模型与拉格朗日建模方法，构建了柔性机器鱼的一体化动力学模型，并采用 Morrison 方程与升阻力模型分析了各部分鱼体所受的水动力。仿真结果验证了所搭建动力学模型的有效性，水下测试结果表明所设计的柔性机器鱼系统能够较好地兼顾游动速度与低能耗目标。

二、针对柔性机器鱼的游速与功耗优化问题，分别提出了面向仿生鱼体波、游速最优与功耗最优的关节刚度优化方法。首先，为了提高所设计机器鱼的仿生特性，提出了一种被动离散拟合方法，通过模拟自然鱼类的鱼体波曲线来获得关节刚度配置，使柔性机器鱼能够较好模仿生物鱼类的运动。在此基础上，为了提高游动速度，提出了一种相位优化方法，将游速优化问题转化为关节相位优化问题，通过优化求解，有效提高了柔性机器鱼的游速。最后，基于动力学理论分析，验证了所设计的柔性关节在降低机器鱼功耗方面的积极作用，并提出了两阶段爬山算法与改进的多种群遗传算法进行功耗最优问题求解，有效降低了柔性机器鱼的功率消耗。经过仿真与实际测试，柔性机器鱼实现了 1.63 BL/s 的最高游速和 4.8 J/m 的最小 COT，具备较好的运动性能。

三、针对柔性机器鱼的转向控制与性能优化问题，提出了一种转向运动控制优化框架。首先，基于运动学分析，设计了三种机器鱼转向控制策略，结果表明不同运动频率下需要调整控制策略以实现较好的性能表现。在此基础上，考虑了柔性机器鱼的非线性动力学特性，基于带约束迭代线性二次规划算法设计了转向性能优化方法，分别在静止状态与稳定运动状态下求解了最优转向控制律，有效提高了柔性机器鱼的静态与动态转向性能。最后，针对柔性机器鱼多性能目标优化问题，提出了一种基于改进 NSGA-II 算法的优化框架，通过引入自适应性记忆空间，有效提高了算法运行效率，并得到了最优 Pareto 前沿，以实现多性能目标的平衡。仿真与水下实验验证了所设计优化框架的有效性。

四、针对柔性机器鱼的路径规划与跟踪控制问题，提出了一种路径优化与控制框架。首先，基于带约束迭代线性二次规划算法进行了局部路径优化器设计，结合 A* 算法的路径导向能力与柔性机器鱼的动力学特性实现了局部避障路 径优化，得到了实数空间上的最优避障路径。在此基础上，设计了局部路径平滑方法，并求解了能耗最优的路径规划问题，得到了总能耗最低的全局路径以及关节刚度配置。最后，针对柔性机器鱼的跟踪控制问题，对柔性机器鱼的头部偏航角进行了周期性平滑，并基于可视距离视线导航法与非线性模型预测控制算法实现了柔性机器鱼的路径跟踪任务，仿真结果验证了所提方法的有效性。

With the advancement and development of science and technology, research scholars have achieved numerous outcomes in the fields of mechanical structure design and motion performance optimization for biomimetic robotic fish, making them increasingly important tools for exploring underwater environments. However, compared to natural fish, robotic fish still exhibit significant disparities in swimming performance and efficiency, necessitating further in-depth research and exploration.

This dissertation focuses on the systematic design and modeling of a biomimetic robotic fish, optimization of swimming performance, design of rapid turning control strategies, and research on path planning and tracking control. The technical contributions are summarized as follows.

Firstly,  in order to achieve high swimming speed and low power cost, a multi-flexible joint robotic fish design scheme is proposed by integrating flexible materials and multi-joint structure, and the prototype is developed. For the planar motion of the robotic fish, an integrated dynamic model is constructed based on the pseudo-rigid body model and the Lagrangian modeling method. The hydrodynamic forces are analyzed using the Morrison equation and the lift and drag force models. Simulation results verify the effectiveness of the developed dynamic model, and aquatic experiments demonstrate that the flexible robotic fish can effectively achieve a balance between swimming speed and energy cost.

Secondly, optimization methods for joint stiffness aiming at mimicking fish body waves, achieving optimal swimming speed, and minimizing power cost are proposed to optimize the swimming speed and power cost of the flexible robotic fish. Considering the biomimetic characteristics of the flexible robotic fish, a passive discrete fitting method is proposed. This method imitates the body wave curves of natural fish to obtain joint stiffness configurations, allowing the flexible robotic fish to effectively replicate the movements of natural fish. By integrating the joint phase into the optimization of swimming speed, a phase optimization approach is proposed to improve the swimming speed of the flexible robotic fish. With regard to power cost, theoretical dynamic analysis substantiates the positive effect of the designed flexible joints in reducing the power cost of the robotic fish. Besides, an optimization framework that combines a two-phase hill-climbing with an improved multiple population genetic algorithm is presented to effectively reduce the power cost of the flexible robotic fish. Through simulations and aquatic tests, the flexible robotic fish achieves a maximum swimming speed of 1.63 BL/s and a minimum COT of 4.8 J/m, demonstrating good locomotive performance.

Thirdly, to address the turning control and performance optimization of the flexible robotic fish, a turning control optimization framework is proposed. Through kinematic analysis, three turning control strategies for the flexible robotic fish are designed. The experimental results indicate that the control strategies need to be adjusted under different motion frequencies to achieve good performance. On this basis, the nonlinear dynamic characteristics of the flexible robotic fish are taken into account, and a turning performance optimization method is designed based on the constrained iterative linear quadratic regulator algorithm. The optimal turning control strategies are solved respectively in static and stable motion states, effectively improving the static and dynamic turning performance of the flexible robotic fish. Considering the multiple performance objectives of the flexible robotic fish, an optimization framework based on an improved NSGA-II algorithm is proposed. By introducing an adaptive memory space, the operational efficiency of the algorithm is effectively improved, and the optimal Pareto frontier is obtained, thus achieving a balance among the multiple performance objectives. The effectiveness of the designed optimization framework is verified through simulation and underwater experiments.

Fourthly, a path optimization and control framework is proposed to address the issues of path planning and tracking control for the flexible robotic fish. A local path optimizer is designed based on the constrained iterative linear quadratic regulator algorithm. By combining the path guidance capability of the A* algorithm with the dynamic characteristics of the flexible robotic fish, local obstacle avoidance path optimization is achieved, resulting in an optimal obstacle avoidance path in real-number space. Besides, a local path smoothing method is designed, and via solving the path planning problem with optimal energy cost, a global path with the lowest energy cost and the corresponding joint stiffness configurations are obtained. Furthermore, the yaw angle of the flexible robotic fish is periodically smoothed, and path tracking is achieved based on the visual distance line-of-sight navigation method and nonlinear model predictive control algorithm. Simulation results demonstrate the effectiveness of the designed framework.