【作者】 王超；

【导师】 陈玲玲；

【作者基本信息】 河北工业大学 ， 工程硕士（专业学位）， 2020， 硕士

【摘要】 助行外骨骼机器人是一种能够给四肢健全的人提供助力输出及机能放大的可穿戴设备,可以显著增强人体机能,因此受到科研人员的广泛关注。在外骨骼系统中,穿戴者与机器应相互协作,相互学习,相互感知,共同思考和决策。本文针对助行下肢外骨骼机器人两关节的轨迹跟随问题展开研究,主要内容如下:首先,根据人体下肢特点考虑一种气压驱动下肢外骨骼机器人,通过合理假设将该可穿戴型下肢外骨骼简化成连杆模型,并对关节角度和气动肌肉收缩状态的关系进行数学分析。基于拉格朗日方法建立下肢外骨骼在摆动相和支撑相的动力学模型,并给出具体模型参数,为后续控制器设计奠定基础。其次,对人体下肢的运动步态进行分析,设计基于Mpu6050的角度采集和基于高动态力传感电阻器的足底信息采集模块,同时实现足底压力数据和关节角度数据的采集、调理和存储等任务,并通过数据的周期性特点对步态周期进行划分。基于MATLAB完成压力数据预处理并据此定义步态特征,采用极限学习机算法,构建步态识别网络,该网络识别结果为控制过程中步态切换提供依据。再次,针对外骨骼机器人系统非线性、不确定性、易受外界干扰等特点,考虑下肢外骨骼状态不可测条件下的步态跟随问题,利用状态反馈信息设计滑模控制器,同时设计扩张状态观测器对系统集总干扰进行估计,补偿滑模控制器以减小其控制增益,削弱抖振现象。针对气动肌肉收气与放气时需要时间以及关节输入力矩在步态切换时不能跳变的特点,设计了主动平滑的切换机制。最后,考虑气压驱动下肢外骨骼气动执行器的特点,将各气动肌肉视为具有“自我计算能力”的智能体,根据气动肌肉物理测试和理论模型,建立多气动肌肉系统。设计基于事件触发的分布式控制器,控制各气动肌肉在保持各自速度的情况下完成分布式控制,多气动肌肉能够到达指定的固定相对位置或动态相对位置,从而驱动外骨骼跟随人体下肢运动。事件触发机制使气动肌肉的控制输入只在“特定时间”更新,减少了气动肌肉之间的信息交换,节省了气动肌肉之间的通讯资源。更多还原

【Abstract】 Assisted exoskeleton robot refers to a wearable device which provides power-boosting and strength-amplification to wearer’s complete limbs.It has been extensively studied because it can significantly enhance human body functions.The wearer and the machine in the exoskeleton system should perform their great work,learn from each other,share perceptions,think and make decisions.This article focuses on the trajectory tracking control of assisted lower extremity exoskeleton robot.The main contents are as follows:First,the study considers a pneumatic-driven lower extremity exoskeleton robot which is based on the characteristics of the human lower limbs.Through reasonable assumptions,the wearable lower extremity exoskeleton structure can be simplified into the two-link model,and then the relationship between joint angle and pneumatic muscle contraction state can be mathematically analyzed.The Lagrangian dynamics method is used to establish the dynamic model of the lower extremity exoskeleton on the swing phase and the stance phase,and the specific system model parameters are given separately.This chapter lays the foundation for the controller designing.Secondly,the human lower limb movement gait is analyzed,an angle acquisition module based on mpu6050 and a plantar pressure acquisition module based on High Dynamic Force Sensing Resistor are designed.The collection,conditioning,and storage of the data are realized simultaneously.The gait period is divided by the periodic characteristics of the data.The data preprocessing is completed based on MATLAB and the gait characteristics are defined according to the processed pressure data.The Extreme Learning Machine Algorithm is used to construct a gait recognition network.The network recognition result provides the basis for gait switching in closed-loop control.Thirdly,aiming at the characteristics of the exoskeleton robot system such as nonlinearity,uncertainty,and susceptibility to external interference,the gait following problem of the extremity exoskeleton under unpredictable conditions are considered,the sliding mode controller and extended state observer are designed by state feedback.The extended state observer estimates the total interference of the system and compensates it to the controller,controller gain and chattering phenomenon of the sliding mode controller are reduced.Due to the "time-consuming" property of pneumatic muscles,the joint input torque cannot jump during gait switching,so an active smooth switching mechanism is designed.Finally,according to the characteristics of the lower extremity exoskeleton pneumatic actuators,each pneumatic muscle actuator is viewed as an agent with "self-computing ability",and a multi-pneumatic-muscle model is established based on pneumatic muscle physical tests and theoretical model.An event-based lower extremity exoskeleton distributed controller is designed,pneumatic muscles maintain distributed control at a certain speed.The pneumatic muscles can reach the specified fixed relative position or dynamic relative position so that the lower extremity exoskeleton shows a coordinated movement with the limb.The control input update time is determined by the event trigger mechanism which reduces information exchange and communication resources between pneumatic muscles.更多还原