【作者】 张代兵；

【导师】 胡德文；

【作者基本信息】 国防科学技术大学 ， 控制科学与工程， 2007， 博士

【摘要】 水下推进技术是决定水中航行器航程、航速和机动性的关键技术,研究新型仿生水下推进器具有重要的经济意义和军事价值。波动鳍仿生水下推进器是一种模仿鱼类中央鳍/对鳍模式(Median and/or Paired Fin,MPF)推进的新型仿生水下推进器,具有机构简单、控制灵活、流体载荷分布均匀等优点,自二十世纪初开始引起国内外研究机构的关注。本文围绕波动鳍仿生水下推进器开展了仿生学、机构设计、控制方法和实验等一系列研究工作,主要研究内容和成果如下:1深入开展了仿生对象——鱼类波动鳍生物推进器的仿生学研究,为波动鳍仿生水下推进器机构设计和控制方法研究提供了客观依据和科学指导。以弓鳍目鱼类“尼罗河魔鬼”的背鳍推进器为仿生对象,基于双视角同步成像观测系统进行实验,获取了仿生对象的形态学特征、内部构造和中枢神经系统控制机理,分析了起动、停止、机动和直线巡游运动过程的动力学特性,建立了波动鳍生物推进器在直线巡游状态下的运动学模型,通过对运动学模型中的波速、波幅、频率三个参数进行调节对鳍面运动形状进行了拟合。2成功设计了并联多关节、多参数可控的波动鳍仿生水下推进器实验装置,深入分析了仿生水下推进器的动力学特性。分析了波动鳍仿生水下推进器的设计原则,研制了多关节并联、直流伺服电机驱动和分层控制的波动鳍仿生水下推进器实验装置;建立了考虑鳍面弹性形变和流体作用的推进器动力学模型,分析了单关节和整个推进器系统的动力学特性,基于作动盘理论建立了波动鳍仿生推进器的简化推力模型。3基于新型神经元振荡器模型设计了一种仿生神经网络控制系统,实现了对波动鳍仿生水下推进器各种运动的有效控制。提出了一种有两个神经元构成、连接关系简单、易于工程应用的新型神经元振荡器模型,其振荡参数由动力学方程中三个系数独立控制。以新型振荡器为基本单元,设计了单个关节的仿生神经网络控制系统,研究了单关节起动、停止、等幅摆动等运动的控制方法。设计了控制波动鳍仿生水下推进器的仿生中枢模式发生器群(Central Pattern Generators,CPGs)神经网络控制系统,研究了推进器起动、停止和稳态游动的控制方法,在引入神经元估计器构成闭环后实现了对推进器波动幅度的稳定控制。实验结果表明:本文设计的仿生神经网络控制方法比传统的逆运动学控制方法更加有效,实现了对仿生对象从“形态相似”到“功能相似”的进步,可应用于类似的仿生机器人。4进行了波动鳍仿生水下推进器实验装置的水动力实验、运动实验和流场实验,分析了推进器运动过程中的推力、侧向力、推进速度和流场变化,对简化推力模型中的参数进行了辨识,分析了目前制约波动鳍仿生水下推进器性能的关键因素。更多还原

【Abstract】 Underwater propulsion technology is an absolutely key component which determines the voyage, velocity and maneuverability of the underwater vehicles. It is full of economic and military significance to develop new-type bionic underwater propulsors for the requirements in better performance, higher efficiency, less disturbance, etc. The underwater bionic undulatory-fin propulsor, inspired by undulations of the median and/or paired fin (MPF) fish, has advantages in simplicity in structure, agility in control and uniformity in fluid load distribution. Such a new-type bionic propulsor has been paid more and more attention to by many related research institutes of the world since 2000s.In this thesis, the bionic inspirations, mechanical structure, control method and experiments of the undulatory-fin propulsor were investigated deepgoingly and systematically. The main research contents are listed as follows:1. Bionic inspirations from the Amiiform fish "Gymnarchus niloticus", which generally swims by undulations of the long dorsal fin, are extracted to answer why and how the mechanical structure and control system are designed for the undulatory-fin propulsor. We established a two-view synchronous imaging apparatus to observe the dorsal fin’s outline as well as shape in the different movements, and to analyze its morphological characteristics, the inner structure as well as the central neural control mechanism, aided by the X-ray images. The swimming characteristics were analyzed during the starting, stopping, straight cruise and maneuvering motion. And furthermore, we set up a kinematical model to describe the line swimming motion of the undulatory-fin propulsor and we can construct several fin shapes by regulating kinematical parameters of the propulsive velocity, amplitude as well as frequency.2. An undulatory-fin experimental device was accomplished which is composed of multiple parallel joints and whose undulation parameters can be independently regulative. Hereby, the dynamic performance was in-depth studied. In detail, we presented three design principles of the bionic undulatory-fin propulsor, built a novel dynamic model with considerations of the elastic fin’s deformation and the unsteady fluid action, and analyzed the dynamic performance of the single joint as well as the whole bionic undulatory device. Moreover, a simply thrust prediction model was set up based on the actuator-disc theory.3. The bionic neural network control system was established with the basis of a new-style neural oscillator, and the various motions of the undulatory-fin propulsor can be effectively controlled by this neural network. The new-style neural oscillator, which is composed of two neurons and whose oscillation parameters are independently controlled by three special coefficients in the dynamic equations, possesses advantages of simple connections and convenient engineering application. Taking the neural oscillator as the basic component, we designned a single-joint neural network control system, and studied its corresponding control algorithms for motions, such as starting, stopping and uniform-amplitude swimming. Furthermore, a bionic CPGs (central pattern generators) neural network control system is presented to control the movements of the undulatory-fin propulsor, and to design the control algorithms for starting, stopping and steady swimming for the undulatory-fin propulsor. What’s more, we accomplished the close-loop control by introducing a neural estimator, and applied this method into controlling the undulatory amplitude. The results verify that the bionic neural network control method can be more effective than the traditional reverse-kinematics method, and can be adaptive to various bionic robots. From the view point of bionics, the work in this thesis may promote the undulatory-fin propulsor from "likeness in shape" to " similarity in spirit".4. We carried out several experiments on hydrodynamics, kinematics and fluid fields, to analyze the thrust, lateral force, swimming velocity and fluid field structure during the undulatory-fin propulsion. To be more meaningful, the parameters in the simply thrust prediction model are identified with the measurement data. And several key factors which have great influence on the undulatory-fin propulsor are proposed and analyzed.更多还原