【作者】 史小华；

【导师】 王洪波； 高峰；

【作者基本信息】 燕山大学 ， 机械电子工程， 2014， 博士

【摘要】 康复机器人是医工技术和机器人技术结合的产物，属于医疗机器人领域。我国人口老龄化的现状日趋严峻，脑卒中或各种意外事故导致的下肢瘫/截瘫患者众多，他们是一个亟需关爱的特殊弱势群体。日益增多的下肢残障患者和严重匮乏的康复医师是现存的主要矛盾。“十二五”期间我国将大力加强康复医疗服务能力的建设。目前上肢康复机器人的研究已经涵盖上肢及手部各关节，下肢康复机器人相对较少，针对当前研究存在的不足，本课题研制了坐/卧式外骨骼型下肢康复机器人，并对其进行了较为系统的研究，主要内容总结如下：阐述了运动神经元构造、肌电信号（Electromyography，EMG）的产生机理及人体下肢骨骼肌的解剖结构，分析了不同肌力等级对应的康复训练方式和方法。基于人体工程学，设计了坐/卧式外骨骼型下肢康复机器人。该机器人包括两条机械腿和靠背角度可调的座椅。可实现被动训练、主动辅助训练和主动抗阻训练三种运动模式。详细剖析了该机器人各部分的机械结构、安全限位保护装置及控制系统硬件构成。简化该机构为三连杆模型，采用“D-H”（Denavit-Hartenberg）法对其运动学进行了理论分析，并用ADAMS软件进行了仿真验证。基于AutoCAD绘图软件采用几何法分析了不同坐/卧姿态下机器人的连续变工作空间，采用构造函数法拟合出了其最大轨迹圆的圆心坐标的轨迹曲线。采用力平衡法，分析了人机系统的静力学正反解，确定了机器人末端力和各关节力的映射关系。采用拉格朗日法（LagrangianMethod）建立了人机动力学模型，分析了其各关节在运动过程中所需的驱动力，用ADAMS软件对其被动训练轨迹为圆时的各关节的运动学和动力学特性进行仿真验证。引入“右腿驱动技术”研制了表面肌电信号（Surface Electromyography，SEMG）采集板，对其滤波、“双T”型陷波、运放等功能模块电路的原理进行了分析和计算，并用EWB电子实验室（Electronics Workbench，现称为MultiSim）软件进行了仿真。试验采集了人体下肢肌的SEMG信号，经过时域内的系列处理，得到其移动平均幅值，和设定的阀值比较可以识别出人的运动意图。根据该机器人控制系统的构成，从应用的角度系统的阐述了三种训练模式的基本思想和可实现的控制策略。建立了单关节的传动系统的传递函数。被动训练采用传统的PID控制，理论推导了其建模过程，并对重力等扰动进行了补偿分析；对多关节被动训练的被控系统进行了线性解耦分析；助力辅助训练采用模糊自适应PID控制策略，并对其结果进行了仿真。单关节主动训练简化为“质量-弹簧”系统模型，对其进行了理论分析。多关节主动训练时采用能实现力追踪的自适应阻抗控制策略，论述了力追踪的自适应算法，并采用Lyapunov稳定性理论对力追踪算法的参数进行了求解。最后对该机器人进行了被动训练试验，并对试验误差进行分析。更多还原

【Abstract】 Rehabilitation robot is a product of the combination of the biomedical engineeringtechnology with robot technology, which belongs to the medical robots. In China, theaging population is more and more, and the state is getting worse. The number of patientswith lower limb dyskinesia or paraplegia caused by stroke or various accidents anddisaster is large. They are special vulnerable groups in need of care urgently. The maincontradiction is between the numbers patients with lower limb paralysis and the shortageof rehabilitation doctors. China will vigorously strengthen construction of therehabilitation medical services capacity during the”Twelfth Five-Year”. Now the researchabout the upper limb rehabilitation robot already covered every joint of upper limb andhand, while the similar research about lower limb rehabilitation robot is less than it.Aimed to the lack of current research, the author and the other members of the teammanufactured an exoskeleton-type lower limb rehabilitation robot and the patient can sitor lying in the seat. This rehabilitation robot is researched systematically, and the maincontents are summarized as follows:The neuron construction, mechanism and the anatomical structure of lower limbskeletal muscle are introduced. The rehabilitation training mode and methods correspondto different muscle force levels are analyzed. The exoskeleton-type lower limbrehabilitation robot is designed based on Ergonomics，and the patients can sit or lying inthe seat. This robot contains two mechanical legs and a seat. The angle of the seat backcan be adjusting. The robot can realized three training modes: passive training mode,assisted training mode and active resistance training mode. The mechanical structure ofthe robot, the hardware of the control system and the protection device are introduced.The robot mechanism is simplified into a three-link model, forward solution andinverse solution of the robot kinematics is analyzed based Denavit-Hartenberg Method,then analysis the theory results and simulation results which is made by ADAMSsoftware.The working space of the robot which can be varied is analyzed based on geometric method with AutoCAD software. Through force-balance method, the statics forwardsolution and inverse solution of human-machine system is analyzed. It identified therelationship between the force belongs to the back end and every joint of the robot. UsingLagrangian method, the human-machine dynamics model is obtained, and the requireddriving force of every joint is analyzed in the process of movement. When the motiontrajectory with passive training mode is circle, the kinematic and dynamics characteristicsare obtained and verified by simulation method using ADAMS software.The circuit principle of SEMG capture board which developed by the authors isintroduced and the composition circuits are simulated and verified using EWB(Electronics Workbench) software. The SEMG signals is collected from lower limbmuscles through experiment, and then obtained the average values after a series ofprocessing. Use it, we can judge the human’s movement intention.It states the basic idea and realizing ways of three training modes. The transferfunction of single joint is set up. The traditional PID control is adopted in the passivetraining mode, and then theoretical derivation of its modeling process. The compensationanalysis is done aimed to the gravity and disturbance. The linear decoupling is made to thecontrolled system in the multi-joint passive training mode. Fuzzy adaptive PID controlstrategy is adapted to the power-assisted training mode, and simulated the process.Single-joint active training is simplified into the mass-spring model, and then analysis it.Adaptive impedance control strategy is adapted to the multi-joint resistance training. Itintroduced the force adaptive algorithms. Then analysis the control system by Lyapunovstability theory. Finally, the passive training experiment was completed by the robot, andanalyzed the experimental error.更多还原