分类号: 密级: UDC: 编号:201232402003 河北工业大学硕士学位论文 自适应性主动式下肢假肢仿生控制 方法研究 论文作者: 苏龙涛 学生类别: 全日制 专业学位类别: 工程硕士 领域名称: 控制工程 指导教师: 郭欣 职 称: 副教授 资助基金项目:国家自然科学基金动力型下肢假肢运动状态识别与协调控制方 法研究61174009和适应复杂环境的主动型假肢建模与平稳控制机理研究61203323。 万方数据
分类号: 密级: U D C: 编号: 201232402003 河北工业大学硕士学位论文 自适应性主动式下肢假肢仿生控制 方法研究 论 文 作 者: 苏龙涛 学 生 类 别: 全日制 专业学位类别: 工程硕士 领 域 名 称: 控制工程 指 导 教 师: 郭欣 职 称: 副教授 资助基金项目:国家自然科学基金动力型下肢假肢运动状态识别与协调控制方 法研究61174009和适应复杂环境的主动型假肢建模与平稳控制机理研究61203323。 万方数据
万方数据
万方数据
Dissertation Submitted to Hebei University of Technology for The Master Degree of Control Engineering RESEARCH ON BIO-INSPIRED CONTROL METHOD OF ADAPTIVE ACTIVE LOWERLIMB PROSTHESES by Su Longtao Supervisor:Prof.GuoXin April 2015 This work is supported by the National Natural Science Foundation of China.No.61174009 and No.61203323. 万方数据
Dissertation Submitted to Hebei University of Technology for The Master Degree of Control Engineering RESEARCH ON BIO-INSPIRED CONTROL METHOD OF ADAPTIVE ACTIVE LOWERLIMB PROSTHESES by Su Longtao Supervisor: Prof. GuoXin April 2015 This work is supported by the National Natural Science Foundation of China. No. 61174009 and No. 61203323. 万方数据
万方数据
万方数据
原创性声明 本人郑重声明:所呈交的学位论文,是本人在导师指导下,进行研究工作所取得 的成果。除文巾已经注明引用的内容外,本学位论文不包含任何他人或集体已经发表 的作品内容,也不包含本人为获得其他学位而使用过的材料。对本论文所涉及的研究 工作做出贡献的其他个人或集体,均已在文中以明确方式标明。本学位论文原创性声 明的法律责任由本人承担。 学位论文作者签名: 苏诗 日期:2019、5、30 关于学位论文版权使用授权的说明 本人完全了解河北工业大学关于收集、保存、使用学位论文的以下规定:学校有 权采用影印、缩印、扫描、数字化或其它手段保存论文,学校有权提供本学位论文全 文或者部分内容的阅览服务;学校有权将学位论文的全部或部分内容编入有关数据库 进行检索、交流:学校有权向国家有关部门或者机构送交论文的复印件和电子版。 (保密的学位论文在解密后适用本授权说明) 学位论文作者签名: 苏龙诗 日期:2015、5,30 师 日期:≥05.5.30 万方数据
万方数据
万方数据
万方数据
摘 要 近些年来,随着生物学和仿生学的研究发展,很多的生物学的研究成果为假肢控 制领域带来新的思路。研究发现,生物的很多运动例如呼吸、行走、心跳等,并不是 由生物体的大脑去直接控制,而是由位于生物低级神经中枢的被称作中枢模式发生器 (Central Pattern Generator,.CPG)去直接控制。本文将基于CPG的仿生控制方法应用 到下肢假肢的控制,主要工作内容包括以下几个方面: 1、熟悉了解人体对步态的分层运动控制系统,将基于CPG的仿生控制方法应用 于对下肢假肢的控制,使用自激振荡的非线性振荡器去模拟控制信号的产生过程,对 常见的Hopf振荡器、Rayleigh振荡器和Matsuoka振荡器进行分析比较,选择Hopf 振荡器作为组成振荡器网络的基本振荡器。 2、Hof振荡器仅能产生固定角频率的振荡信号,无法实现对外部训练信号的学 习,因此将Dynamic Hebbian Learning学习算法加入到对Hopf振荡器的学习训练之中, 改进的Hof振荡器中的角频率由常量变为状态变量,进而可以实现对外部周期信号 角频率的学习。 3、将几个改进的Hopf振荡器组成CPG网络,振荡器之间实现耦合,使用正常 的肢体行走步态数据去完成对整个网络的训练,并对整个振荡器网络引入反馈,用反 馈信息去调整每个振荡器的幅值、振荡角频率和相位。训练结束后,由若干具有不同 幅值、振荡角频率和相位的振荡器组成的CPG网络,可以实现对训练信号的重现, 振荡器网络自激振荡输出可以用于控制假肢的信号,步速的调整和步态的切换。 4、使用双下肢仿生平台作为测试算法的平台,使用Labview搭建上位机,实验 结果证明可以将基于CPG的仿生控制方法应用于下肢假肢的控制,CPG产生的假肢 控制信号和人体产生的用于肢体控制的信号十分接近,方法是可行的。 关键词:Hopf振荡器中枢模式发生器主动式假肢仿生控制方法 I 万方数据
I 摘 要 近些年来,随着生物学和仿生学的研究发展,很多的生物学的研究成果为假肢控 制领域带来新的思路。研究发现,生物的很多运动例如呼吸、行走、心跳等,并不是 由生物体的大脑去直接控制,而是由位于生物低级神经中枢的被称作中枢模式发生器 (Central Pattern Generator, CPG)去直接控制。本文将基于 CPG 的仿生控制方法应用 到下肢假肢的控制,主要工作内容包括以下几个方面: 1、熟悉了解人体对步态的分层运动控制系统,将基于 CPG 的仿生控制方法应用 于对下肢假肢的控制,使用自激振荡的非线性振荡器去模拟控制信号的产生过程,对 常见的 Hopf 振荡器、Rayleigh 振荡器和 Matsuoka 振荡器进行分析比较,选择 Hopf 振荡器作为组成振荡器网络的基本振荡器。 2、Hopf 振荡器仅能产生固定角频率的振荡信号,无法实现对外部训练信号的学 习,因此将Dynamic Hebbian Learning学习算法加入到对Hopf振荡器的学习训练之中, 改进的 Hopf 振荡器中的角频率由常量变为状态变量,进而可以实现对外部周期信号 角频率的学习。 3、将几个改进的 Hopf 振荡器组成 CPG 网络,振荡器之间实现耦合,使用正常 的肢体行走步态数据去完成对整个网络的训练,并对整个振荡器网络引入反馈,用反 馈信息去调整每个振荡器的幅值、振荡角频率和相位。训练结束后,由若干具有不同 幅值、振荡角频率和相位的振荡器组成的 CPG 网络,可以实现对训练信号的重现, 振荡器网络自激振荡输出可以用于控制假肢的信号,步速的调整和步态的切换。 4、使用双下肢仿生平台作为测试算法的平台,使用 Labview 搭建上位机,实验 结果证明可以将基于 CPG 的仿生控制方法应用于下肢假肢的控制,CPG 产生的假肢 控制信号和人体产生的用于肢体控制的信号十分接近,方法是可行的。 关键词:Hopf 振荡器 中枢模式发生器 主动式假肢 仿生控制方法 万方数据
ABSTRACT Referencing to the research results in biological and neurology fields in recent years, many new ideas are put forward to control the prosthesis.The study find that many movement such as respiratory,gait walking,heartbeat,etc.,are not controlled directly by the organism's brain,instead are controlled directly by the Central Pattern Generator(CPG), which is located in lower nerve center.In this article,the biological control method based on CPG is used to control the prosthesis.The main contents of could be presented as follows: 1.By introducing the layered motion control system of human,then the bio-inspired control method based on CPG is make used to the control of lower limb prosthetic,and the self-excited oscillator is used to simulate the signal produced by the neuron,which can be used to control the prosthesis.There are several common oscillator models such as Hopf oscillator,the Rayleigh oscillator and the Matsuoka oscillator.After compared with each other,the Hopf oscillator is selected as the basic unit of the CPG network. 2.Because the self-excited Hopf oscillator can only generate the signal with fixed frequency,and cannot adapt to the external training signal,until introduce the Dynamic Hebbian Learning algorithm to the training,in which the frequency of the training signal can be learned based on the change of angular frequency from a constant to a state variable. 3.By constructing the CPG network with several improved and self-adapted Hopf oscillators,the coupling model is achieved between the oscillators.Introducing feedback to the CPG oscillatory network can be used to adjust the value of the amplitude,the angular frequency and the phase of the oscillators.The normal lower limb gait information is used to train the oscillatory network.After that,the oscillatory network with the oscillators that have different amplitude,angular frequency and phase,can reproduce the training signal, which can be used to control the AK prosthsis. 4.A prosthetic experimental platform was used in Labview to validate the simulation results.The output from CPG controller for AK prosthesis is very close to the joint trajectory of human body.The experiment result prove that use the bionic method based on CPG to control the prosthetic is feasible. II 万方数据
II ABSTRACT Referencing to the research results in biological and neurology fields in recent years, many new ideas are put forward to control the prosthesis. The study find that many movement such as respiratory, gait walking, heartbeat, etc., are not controlled directly by the organism's brain, instead are controlled directly by the Central Pattern Generator (CPG), which is located in lower nerve center. In this article, the biological control method based on CPG is used to control the prosthesis. The main contents of could be presented as follows: 1. By introducing the layered motion control system of human, then the bio-inspired control method based on CPG is make used to the control of lower limb prosthetic, and the self-excited oscillator is used to simulate the signal produced by the neuron, which can be used to control the prosthesis. There are several common oscillator models such as Hopf oscillator, the Rayleigh oscillator and the Matsuoka oscillator. After compared with each other, the Hopf oscillator is selected as the basic unit of the CPG network. 2. Because the self-excited Hopf oscillator can only generate the signal with fixed frequency, and cannot adapt to the external training signal, until introduce the Dynamic Hebbian Learning algorithm to the training, in which the frequency of the training signal can be learned based on the change of angular frequency from a constant to a state variable. 3. By constructing the CPG network with several improved and self-adapted Hopf oscillators, the coupling model is achieved between the oscillators. Introducing feedback to the CPG oscillatory network can be used to adjust the value of the amplitude, the angular frequency and the phase of the oscillators. The normal lower limb gait information is used to train the oscillatory network. After that, the oscillatory network with the oscillators that have different amplitude, angular frequency and phase, can reproduce the training signal, which can be used to control the AK prosthsis. 4. A prosthetic experimental platform was used in Labview to validate the simulation results. The output from CPG controller for AK prosthesis is very close to the joint trajectory of human body. The experiment result prove that use the bionic method based on CPG to control the prosthetic is feasible. 万方数据
KEYWORDS:Hopf Oscillator Central Pattern Generator Active ProsthesisBionic Control Method III 万方数据
III KEYWORDS: Hopf Oscillator Central Pattern Generator Active ProsthesisBionic Control Method 万方数据
IV 万方数据
IV 万方数据