【作者】 黄鹏程；

【导师】 张立彬； 杨庆华；

【作者基本信息】 浙江工业大学 ， 机械电子工程， 2013， 博士

【摘要】 仿生康复是现阶段康复医学及康复器械研究领域的核心思想。传统的康复器械,尤其是针对手部的康复器械,存在结构简单、自由度少、控制策略单一、运动模式少以及适应性、灵活性较弱等不足之处,限制了手部康复器的进一步发展和深入研究。本文设计的仿生康复手是一种模拟人手布局、结构、尺寸,具有多自由度、多种弯曲关节的,采用多点连续肌电控制的新型手部康复器,其结构的灵活性和适应性,控制策略的连续性仿生性,积极推动手部康复器的研究和发展,具有研究意义与应用价值。随着科技的发展和患者需求的提高,依靠电机驱动刚性结构的传统康复手在研究和应用两方面都出现了瓶颈。仿生康复概念的提出,为手部康复器的研究提供了新的理论基础,指出新的研究方向。本文总结了现有康复手的优缺点,在此基础上提出一种气体驱动刚柔性结构的新型仿生康复手,具有良好的被动柔性,同时在控制策略方面弥补了现有康复手的不足,提高了仿生效果。本文主要完成的研究工作如下：(1)根据人类手指关节运动特性对关节弯曲性能的不同要求及关节结构大小的限制,设计具有两种类型弯曲关节、两个自由度的仿生康复手单指基本结构,和具有五个手指、十个自由度的仿生康复手本体结构。以人手布局和实际尺寸为标准,优化仿生康复手的整体布局方案和结构参数。在此基础上,设计仿生康复手的总体控制方案。总结了本文设计的仿生康复手的特色之处。(2)设计了采用双气动柔性驱动器驱动的、模拟人手指关节弯曲运动的新型气动弯曲关节。基于静力学和弹性力学理论,对关节FPA的活动端进行力平衡分析,建立关节转角的静态模型,分析关节结构参数对其弯曲性能的影响；对静态模型进行了简化,并验证了其与理论模型的误差；在简化模型的基础上,根据热力学第一定律和关节动力学方程,推导了关节的转角动态方程组；仿真分析了关节结构参数与弯曲角度之间的关系。实验结果表明,当FPA内腔气压在0.1MPa～0.35MPa时,实际弯曲角度与理论值偏差较小；随着FPA内腔气压增大,偏差也随之增大；当达到理论最大气压时,偏差值为7.5。,偏差率为6.8%；因此可以认为,弯曲关节的转角静态特性与理论模型基本吻合。(3)基于稀疏理论,提出表面肌电信号的特征重构算法。在正交匹配算法的基础上,结合以K-SVD算法构造的自适应原子库,对表面肌电信信号进行稀疏分解。根据稀疏参数与稀疏结果之间的关系,优化稀疏结果。在特征重构算法的基础上,再次提取稀疏结果的特征参数并对其进行重构。对比特征参数曲线与手指关节实际弯曲角度,验证与人手实际角度变化规律的符合度,为后续研究提供理论基础；通过与传统平均幅值算法处理结果的对比,验证本章提出的特征重构算法的优越性,并分析传统算法处理效果欠佳的原因。(4)基于幅值乘方理论,提出表面肌电信号的预处理算法。在表面肌电信号集中参数模型的基础上,分析算法的理论依据。根据人工神经网络理论和误差反向传播算法,设计了用于表面肌电信号模式识别的神经网络分类器。对比预处理前后信号特征参数的分类器训练结果及其识别率,验证预处理算法对信号特征的加权效果。在预处理算法的基础上,对第四章的特征参数曲线进行了修正。(5)在传统肌电控制理论的基础上,根据仿生康复手的整体布局和结构特点,提出多点连续肌电控制方法,模拟人手的运动规律,详细阐述了两者的联系和区别；详细分析表面肌电信号特征参数与仿生康复手控制信号之间的映射关系,并提出其数学模型。在MATLAB和DELPHI编译环境下,设计仿生康复手的控制系统。搭建实验平台,完成仿生康复手的系统控制效果实验。实验结果表明：系统可以较好地识别出信号的动作部位,其整体识别率为96.33%,总体控制误差稳定在5%之内,控制点的平均误差值约为0.5087°,最大偏差控制在1°之内。本文设计的仿生康复手,采用本课题组自主研发的气动柔性驱动器FPA直接驱动,其结构简单、便于控制、易于实现整体结构的小型化、具有良好的被动柔性,同时仿生康复手的刚性结构保证了其整体刚度。仿生康复手的多点连续肌电控制策略使其比传统康复手更符合人手的自然运动规律和动作习惯,更能体现仿生康复的概念。多点连续肌电控制策略不仅仅可以应用于康复器械领域,对于实现其他类型机器人(如工业生产机器人,农业采摘机器人等)的远程操作、仿生操作也有积极的指导意义。更多还原

【Abstract】 The bionic rehabilitation is core ideology in the field of rehabilitation medicine and equipment at this stage. Because of simple structure, fewer DOFs, singleness control strategy, less movement patterns, weak adaptability, poor flexibility and some other shortcomings, the development and application of traditional rehabilitation equipment is severely limited, especially for hand rehabilitation equipment. Simulating the arrangement, structure and size of human hand, bionic rehabilitation hand designed in this paper with multiple DOFs, multiple joints and multi-point continuous myoelectric control is a new type of hand rehabilitation equipment. And its flexible and adaptable structure, as well as its consecutive and biomimetic control strategy plays a positive role to promote research and development of hand rehabilitation equipment. Therefore bionic rehabilitation hand has both great research significance and application value.With the development of science and technology and the increment of patients’needs, the traditional rehabilitation hand with rigid structure driven by motor has turned up bottlenecks in both aspects of research and application. And bionic rehabilitation provides a new theoretical foundation and points out a new research direction of rehabilitation hand. Base on the shortcomings of existing rehabilitation hand summarized in this paper, a new type of bionic rehabilitation hand with rigid and flexible structure driven by gas is designed. It has good passive flexibility, makes up for the deficiencies of the existing rehabilitation hand in the aspect of control strategy and improves the bionic effect of rehabilitation hand.The main research work in this paper is as follows:(1) According to different requirement on the bending property of motion characteristics of human finger joints and limit to the structure size of joint, the single finger basic structure and the body structure of bionic rehabilitation hand have been designed. The single finger basic structure has two joints and two DOFs, while the body structure of bionic rehabilitation hand is designed with five fingers and ten DOFs. The structural parameters and layout scheme are optimized based on layout and actual size of human hand. On this basis, overall control scheme is designed. And characteristic of bionic rehabilitation hand designed in this paper is summarized.(2) A new type of pneumatic bending joint driven by double FPAs is designed to simulate bending movement of human hand. Base on the theory of statics and elasticity, force equilibrium equation of the FPA free end is analyzed. The static model of joint’s bending angle is obtained to analyze the influence of structural parameters on the flexural properties of joint. The simplified model is proposed and its error is analyzed. According to the first law of thermodynamics and the joint’s dynamic equation, joint’s angle dynamic equation is performed. Simulation is carried out to analyze the relationship between structural parameters and bending angle of joint. The result show:when inner pressure of FPA is in0.1Mpa～0.35Mpa, the deviation between actual bending angle and its theoretical value is smaller. With the increment of inner pressure, the deviation increases. When pressure reaches the theoretical maximum value, its corresponding deviation is7.5°and deviation rate is6.8%. Therefore, it can be considered that static characteristics agree with its theoretical model.(3) The feature reconstruction algorithm of sEMG signal is proposed based on the sparse theory. According to orthogonal matching pursuit algorithm, and combined with the adaptive atoms library structured by K-SVD, sEMG signal is decomposed. The relationship between sparse parameters and sparse result is analyzed to optimize the sparse result. Based on feature reconstruction algorithm, the feature of sparse result is extracted and reconstructed. The feature curve is compared with the actual bending angle of joint to verify the conformity of the feature curve, and to provide the theoretical basis for follow-up research. The advantage of feature reconstruction algorithm proposed in this paper can be proved by contrasting with the traditional average amplitude algorithm. And the reason of poor effect of traditional algorithm is analyzed.(4) Based on amplitude power theory, the preprocessing algorithm of sEMG signal is proposed. According to lumped parameter model of sEMG signal, theory evidence of preprocessing algorithm is analyzed. BP neural network classifier is designed for pattern recognition of sEMG signal base on artificial neural network theory and error back-propagation algorithm. The weighted effect of the preprocessing algorithm is proved by contrasting the training results and recognition rate of classifier. And the feature curve of sEMG signal is fixed based on the algorithm proposed.(5) According to characteristics of overall layout and structure of bionic rehabilitation hand, the multi-point continuous myoelectric control method is proposed based on traditional theory to imitate human movement rule. The connection and difference between two theories is elaborated. The mathematical model is deduced based on analyze of mapping relationship between feature parameters of sEMG signal and control signals of bionic rehabilitation hand. The control system is designed in the MATLAB and DELPHI complier environment. Experimental platform is built for experiment of bionic rehabilitation hand control system. The experimental results show:the system can identify the different action parts of the signal. The overall recognition rate is96.33%, while the overall control error is within5%. The average error value of control points is about0.5087°, and the maximum deviation is less than1°.A new type of bionic rehabilitation hand designed in this paper is directly driven by flexible pneumatic actuator developed by our research team. FPA has characteristics of simple structure, easy control, and easy miniaturization of overall structure and so on. It has better passive flexibility. Meanwhile, overall stiffness is ensured by rigid structure of the hand. Compared with the traditional strategy, multi-point continuous myoelectric control strategy is more in line with human hand movement rule. And this strategy not only can be applied to rehabilitation equipment, but also has positive significance for remote operation and bionic operation of other type of robots, such as industrial robot, agricultural picking robot.更多还原