基于气动肌肉仿人下肢控制系统研究

发布时间：2018-01-19 07:17

本文关键词： 仿人下肢气动肌肉并联机构结构矩阵运动学虚功原理动力学模糊自适应控制神经网络　出处：《浙江大学》2017年博士论文　论文类型：学位论文

【摘要】：气动肌肉作为一种由智能材料构成的气动执行元件,具有类似于人类肌肉的柔顺性的特点,同时人肢体的肌肉错综复杂由内、中、外多层肌肉协调实现屈伸、收展、内外旋,因此气动肌肉在仿生机器人领域具有广阔的应用前景。为了了解人的下肢在运动过程中肌肉和关节特性,首先,设计了基于气动肌肉的仿人下肢的机械结构和控制系统,简化了设计的机构并且进行分析和优化;其次,测试了气动肌肉的特性和推导了不规则分布肌肉驱动关节的运动学、动力学方程;最后,提出了基于观测器的神经网络模糊自适应控制策略,并且通过实验进行验证。论文分六章进行阐述,各章主要内容如下:第一章,概述了气动肌肉发展历程和驱动方式,总结了气动肌肉静态模型和动态模型、智能控制策略的国内外研究状况,阐述了气动在仿生领域的研究进展,介绍了并联机构的分类、应用、控制策略,最后概述了本课题的研究意义和研究内容。第二章,基于人体下肢骨骼肌的基本原理,设计了长度不等气动肌肉驱动不规则分布和踝关节环转运动转移到膝关节的多关节并联机构仿人下肢、Field-Programmable Gate Array(FPGA)输出多路Pulse Width Modulation(PWM)的软件电路和高速开关阀的驱动放大硬件电路,选择了气动肌肉、高速开关阀等关键元器件,最后搭建了仿人下肢的控制系统。第三章,推导了关节肌肉运动学正反解和关节的结构矩阵,仿真分析了肌肉的位移、速度和加速度特性,优化了仿人下肢的机械结构。第四章,搭建了气动肌肉测试试验台,并测试了其力学输出特性。分析了关节力矩输出和惯量辨识,提出了肌肉简化为单缸体执行机构与系统的结构矩阵、虚功原理结合求解动力学方程,最终转化为状态方程。第五章,角度传感器作为整体反馈传感器替代传统的每根肌肉并联安装拉线式传感器测量其长度变化,提出了模糊自适应整体控制髋关节的方法,采用神经网络的梯度下降法优化计算高斯基模糊控制的参数,引入鲁棒项以保证系统的鲁棒性,并引入观测器观测不能够直接测量的角加速度,设计了自适应率并用李雅普诺夫函数验证了系统的稳定性。最后,通过实验研究髋关节步态规划和轨迹跟踪过程中气动肌肉、关节运动学和动力学特性。第六章对本论文的研究方法、研究内容、研究结论和创新点进行了总结,并且展望了本课题未来的研究工作。
[Abstract]:Pneumatic muscle, as a pneumatic actuator composed of intelligent materials, is similar to the flexibility of human muscles. At the same time, the muscles of human limbs are intricate and complex to achieve flexion and extension by internal, middle and outer layers of muscles. In order to understand the characteristics of muscles and joints of human lower limbs in the course of movement, pneumatic muscles have a broad application prospect in the field of bionic robots. The mechanical structure and control system of humanoid lower extremity based on pneumatic muscle are designed, the design mechanism is simplified and analyzed and optimized. Secondly, the characteristics of pneumatic muscles are tested and the kinematics and dynamics equations of irregularly distributed muscle-driven joints are derived. Finally, a fuzzy adaptive control strategy based on observer neural network is proposed and verified by experiments. The thesis is divided into six chapters, the main contents of each chapter are as follows: chapter 1. The development history and driving mode of pneumatic muscle are summarized, the static and dynamic models of pneumatic muscle, the research status of intelligent control strategy at home and abroad are summarized, and the research progress in biomimetic field of pneumatic is described. The classification, application and control strategy of parallel mechanism are introduced. Finally, the research significance and research content of this subject are summarized. Chapter 2, based on the basic principle of human lower limb skeletal muscle. A multi-joint parallel mechanism driven by irregular distribution and rotation of ankle joint to the knee joint is designed. Field-Programmable Gate Arraya FPGA) outputs multichannel Pulse Width Modulation (PWM). The software circuit and the drive and amplify hardware circuit of the high speed switch valve. The pneumatic muscle, high speed switch valve and other key components are selected. Finally, the control system of humanoid lower limb is built. In chapter 3, the forward and negative kinematics solutions of joint muscle and the structure matrix of joint are derived. The displacement, velocity and acceleration characteristics of the muscle are simulated and analyzed, and the mechanical structure of the humanoid lower limb is optimized. Chapter 4th, a pneumatic muscle test rig is built. The mechanical output characteristics are tested, the joint torque output and inertia identification are analyzed, and the structure matrix of the single cylinder actuator and system is proposed. The virtual work principle is combined to solve the dynamic equation. Finally converted to the equation of state. 5th chapter, angle sensor as an integral feedback sensor instead of the traditional parallel installation of each muscle pull-wire sensor to measure its length change. A fuzzy adaptive global control method for hip joint is proposed. The gradient descent method of neural network is used to optimize the parameters of Gowski fuzzy control. The robust term is introduced to ensure the robustness of the system. The adaptive rate is designed and the stability of the system is verified by Lyapunov function. The kinematics and dynamics of pneumatic muscles and joints in the process of gait planning and trajectory tracking of hip joint are studied experimentally. Chapter 6th summarizes the research methods, contents, conclusions and innovations of this paper. And the future research work of this subject is prospected.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TP273

【相似文献】