en
×

分享给微信好友或者朋友圈

使用微信“扫一扫”功能。
通讯作者:

张晓旭,E-mail:zhangxiaoxu@fudan.edu.cn

中图分类号:TP242

文献标识码:A

文章编号:1672-6553-2024-22(2)-068-009

DOI:10.6052/1672-6553-2023-031

参考文献 1
康艳梅.2010 年中国残疾人事业发展统计公报[R].北京:中国残疾人联合会,2010. KANG Y M,.2010 China disability development statistics bulletin [R].Beijing:China Disabled Persons' Federation,2010.(in Chinese)
参考文献 2
王富军.中国糖尿病足防治指南(2019版)解读[J].河北医科大学学报,2019,40(11):1241-1245+1250.
参考文献 3
ZIEGLER-GRAHAM K,MACKENZIE E J,EPHRAIM P L,et al.Estimating the prevalence of limb loss in the United States:2005 to 2050 [J].Archives of Physical Medicine and Rehabilitation,2008,89(3):422-429.
参考文献 4
MILLER W C,DEATHE A B,SPEECHLEY M,et al.The influence of falling,fear of falling,and balance confidence on prosthetic mobility and social activity among individuals with a lower extremity amputation [J].Archives of Physical Medicine and Rehabilitation,2001,82(9):1238-1244.
参考文献 5
GABERT L,TRAN M,LENZI T.Design of an underactuated powered ankle and toe prosthesis [C]//2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society(EMBC).New York,USA:IEEE,2021:4920-4923.
参考文献 6
AZOCAR A F,MOONEY L M,DUVAL J F,et al.Design and clinical implementation of an open-source bionic leg [J].Nature Biomedical Engineering,2020,4:941-953.
参考文献 7
ELERY T,REZAZADEH S,NESLER C,et al.Design and validation of a powered knee-ankle prosthesis with high-torque,low-impedance actuators [J].IEEE Transactions on Robotics,2020,36(6):1649-1668.
参考文献 8
WANG X,ZHANG Y,LIANG W,et al.Design,control,and validation of a polycentric hybrid knee prosthesis [J].IEEE Transactions on Industrial Electronics,2023,70(9):9203-9214.
参考文献 9
ROUSE E J,VILLAGARAY-CARSKI N C,EMERSON R W,et al.Design and testing of a bionic dancing prosthesis [J].PLoS One,2015,10(8):e0135148.
参考文献 10
BHAKTA K,CAMARGO J,KUNAPULI P,et al.Impedance control strategies for enhancing sloped and level walking capabilities for individuals with transfemoral amputation using a powered multi-joint prosthesis [J].Military Medicine,2020,185:490-499.
参考文献 11
MA X,XU J,ZHANG X X.Bilateral constrained control for prosthesis walking on stochastically uneven terrain [J].International Journal of Mechanical Sciences,2023,239:107896.
参考文献 12
URREA C,PASCAL J.Design,simulation,comparison and evaluation of parameter identification methods for an industrial robot [J].Computers and Electrical Engineering,2018,67:791-806.
参考文献 13
LI Y B,GUAN X Y,LI W,et al.Dynamic parameter identification of a human-exoskeleton system with the motor torque data [J].IEEE Transactions on Medical Robotics and Bionics,2022,4(1):206-218.
参考文献 14
DALLALI H,FICANHA E,RASTGAAR AAGAAH M.Dynamic modeling of a 2-DOF cable driven powered ankle-foot prosthesis [C]//Proceedings of ASME 2016 Dynamic Systems and Control Conference.New York,USA:American Society of Mechanical Engineers,2017.
参考文献 15
YANG T,WU F,LIU M,et al.An optimization-based approach for prosthesis dynamic modeling and parameter identification [C]//Proceedings of ASME 2015 Dynamic Systems and Control Conference.New York,USA:American Society of Mechanical Engineers,2016.
参考文献 16
ISMAIL M A A,WINDELBERG J,LIU G J.Simplified sensorless torque estimation method for harmonic drive based electro-mechanical actuator [J].IEEE Robotics and Automation Letters,2021,6(2):835-840.
参考文献 17
BARTLETT H L,KING S T,GOLDFARB M,et al.Design and assist-as-needed control of a lightly powered prosthetic knee [J].IEEE Transactions on Medical Robotics and Bionics,2022,4(2):490-501.
参考文献 18
TRAN M,GABERT L,HOOD S,et al.A lightweight robotic leg prosthesis replicating the biomechanics of the knee,ankle,and toe joint [J].Science Robotics,2022,7(72):eabo3996.
参考文献 19
BROCKETT C L,CHAPMAN G J.Biomechanics of the ankle [J].Orthopaedics and Trauma,2016,30(3):232-238.
参考文献 20
WANG S Y,DIAO B B,ZHANG X X,et al.Adaptive signal-correction-based identification for friction perception of the vibration-driven limbless robot [J].Nonlinear Dynamics,2022,108(4):3817-3837.
参考文献 21
CHEN P J,GAO D L,WANG Z H,et al.Study on multi-segment friction factors inversion in extended-reach well based on an enhanced PSO model [J].Journal of Natural Gas Science and Engineering,2015,27:1780-1787.
参考文献 22
ZHANG X X,JI J C,FU J S,et al.Denoising identification for nonlinear systems with distorted streaming [J].International Journal of Non-Linear Mechanics,2019,117:103224.
目录contents

    摘要

    基于模型的智能假肢控制方式具有物理意义明确、参数变量少的优势,但受建模误差、模型不确定等因素的影响,其控制精度仍有待进一步提高,而有效措施之一便是假肢先验动力学建模与辨识.本文针对实验室新设计的动力大腿假肢,研究了基于库伦-粘性摩擦的大腿假肢动力学参数辨识问题.首先,基于拉格朗日方法建立了具有固定传动比动力膝关节和非线性传动比动力踝关节的大腿假肢动力学模型;其次,采用库伦-粘性摩擦模型来描述假肢动力学模型中的关节摩擦行为;最后,通过粒子群优化算法辨识了大腿假肢的动力学参数.结果表明,相比于基于3D建模软件的估计参数,基于辨识参数重构的膝、踝关节电机扭矩与实测扭矩的均方根误差分别降低了93.55%和80.83%,模型精度得到了显著提高.这一结果不仅验证了本文假肢动力学建模与参数辨识方法的有效性,也为假肢后续的高精度控制提供了技术支撑.

    Abstract

    Model-based prosthetic control strategies are still incompetent for practical implementations because modeling uncertainties and errors limit their precision. This paper investigates identifying the dynamic parameters of ankle-knee prostheses based on Coulomb-viscous friction for a newly designed powered prosthesis in the laboratory. Firstly, a powered ankle-knee prosthesis dynamic model, considering the fixed transmission ratio for the knee joint and the nonlinear transmission ratio for the ankle joint, is developed based on the Lagrange method. Secondly, a Coulomb-viscous friction model is employed to describe the joint friction characteristics in the prosthesis dynamic model. Finally, the dynamic parameters of the powered prosthesis are identified by the particle swarm optimization algorithm. Compared with CAD estimation, the root-mean-square errors between the actual and the reconstructed torques after parameter identification are reduced by 99.07% for the knee joint and 83.33% for the ankle joint, indicating that the model precision has been significantly enhanced. The effectiveness, as a consequence, provides a solid technical foundation for accurate prosthesis control.

  • 引言

  • 肢体的缺失给人们正常生活带来了诸多困难,特别是对于那些下肢肢体发生损伤的人,行走能力丧失严重限制了截肢者的活动空间和正常生活能力.据统计,我国现有下肢残疾约2412万人[1],其中约11.4%需要安装假肢.另外,由于每年肥胖症和糖尿病发病率的增加,糖尿病足溃疡截肢者将迅速增加[23].下肢截肢者恢复行走能力的重要方式是穿戴合适的假肢.绝大多数截肢者穿戴传统被动假肢,无法为关节提供足够的输入力矩以保证基本行走步态的协调,这导致每年大约75%的下肢截肢患者二次摔倒[4]和其它诸多慢性健康问题.

  • 相比于被动假肢,动力假肢因可以在截肢者日常行走时提供输入净功,从而可以更好地恢复截肢者行走的生物力学功能[5].在过去十年的动力假肢研究中,动力假肢设计和控制所实现的各种驱动方案主要分为两类,一类是采用同步带、齿轮等具有固定传动比的减速驱动方案[67];另一类则是采用滚珠丝杠或连杆结构的传动驱动装置[89].滚珠丝杠或连杆结构的传动驱动装置虽然可以使得关节驱动电机尽可能远离下肢关节,进而减小假肢的摆动惯量,提升假肢响应的灵敏度.但这些驱动方案往往给假肢动力系统带来高阻抗、大的摩擦损耗和大的反射惯性.这对明确假肢动力学特性和摩擦行为以提高假肢控制性能提出了更高的要求.

  • 动力假肢在控制上往往采用阻抗控制策略[10],其把动力假肢关节简化为虚拟的刚度和阻尼,从而易实现关节扭矩的顺应性控制.然而阻抗控制在步态上控制不连续、也不能实现精确的位置或力控制.动力假肢面临的主要挑战之一是为截肢者提供在户外不同地面条件(包括沙地、草地等非硬质地面等)上的行走能力.相比于传统的阻抗控制,一些基于模型的控制方法有望在动力下肢假肢上实现不同地形上的稳定行走[11].然而,这些基于模型的假肢控制器的性能受限于建模误差、模型参数的不确定性和非线性摩擦等因素的影响[12].明确假肢的动力学模型有助于动力假肢的设计、模拟和优化,有助于开发基于模型的高性能控制器,如逆动力学控制[13].

  • 文献调研表明针对动力假肢参数辨识的相关研究还比较少.Dallali[14]对一绳驱动踝足假肢模型进行了辨识.Yang[15]基于能量方法建立了单膝假肢理想动力学模型.然而针对动力膝-踝假肢的动力学参数识别还没有发现相关文献.在工业机器人或腿式机器人领域,机器人的动力学参数通常是利用外部扭矩传感器测量实验值与仿真计算结果的二次误差来确定的[16].然而,在假肢领域,考虑到重量和成本,一般在关节处不会具有很精确的扭矩测量传感器.此外,关节扭矩传感器也不能完全用于识别关节执行器内部的动力学参数.对于动力假肢,驱动装置除了使用常用的固定传动比的传动结构外,通常使用连杆结构[1718].这些连杆结构的传动比通常是强非线性的,这给辨识工作带来了困难.

  • 本文针对实验室新设计的具有固定传动比的膝关节和具有非线性传动比的踝关节的大腿假肢进行动力学参数辨识.通过建立假肢的动力学模型,实现了对动力膝踝假肢的系统性参数识别.本文提供了一种识别大腿假肢动力学的系统性方法,验证了库伦-粘性摩擦模型在表征假肢关节摩擦行为的性能,提高了假肢动力学模型参数的准确性.

  • 1 动力大腿假肢结构

  • 图1 大腿假肢设计:(a)效果图;(b)实物图

  • Fig.1 Ankle-knee prosthesis design: (a) schematic view; (b) physical picture

  • 图1为实验室设计的动力大腿假肢实物图,由动力膝关节和动力踝关节组成.其中,动力膝关节的电机通过谐波减速器把动力输出给膝关节,编码器安装在电机转子上.动力踝关节的电机通过滚珠丝杠和连杆结构把动力输出给踝关节,其中一个相对编码器安装在电机转子上,另外一个绝对编码器则安装在踝关节处.

  • 2 大腿假肢动力学建模

  • 大腿假肢动力关节通常设计紧凑,其动力学参数辨识时很难通过在关节外部加装扭矩传感器来测量关节扭矩.因此,需要通过电机电流来估计关节扭矩.而电机到关节处的运动经过了减速机构,因此减速机构的传动比是动力学中的一个重要参数.

  • 对于图1所示的膝关节,减速机构为谐波减速器,其传动比为51;对于踝关节,减速机构为滚珠丝杠加连杆结构,其减速比与设计的踝关节传动机构相关,因此,需要通过踝关节传动机构的几何关系计算出踝关节的传动比.

  • 2.1 踝关节非线性传动比计算

  • 如图2所示,踝关节电机转动先经过滚珠丝杠转化为螺母的上下平动,然后经过连杆结构转化为踝关节摆动.

  • 图2 踝关节传动结构

  • Fig.2 Schematic view of the ankle joint transmission structure

  • 根据滚珠丝杠螺母运动的距离与电机转速关系可得:

  • Δd=pθma2π
    (1)
  • 式中,Δd为丝杠移动距离,θma为踝关节驱动电机转角,p为丝杠螺距.

  • 滚珠丝杠螺母运动Δd后,连杆CO的长度变为:

  • CO¯=CO¯0+Δd
    (2)
  • 式中,CO¯0表示小腿和足之间的相对转角θa为零时的连杆初始长度.

  • AC垂直于CO可得:

  • AO¯=CO¯2+AC¯2
    (3)
  • 式中,CO¯AO¯CO¯AC¯分别为图示中各连杆长度.

  • OD¯OC¯共线,可得:

  • BOA+COA+BOF+θa=π
    (4)
  • 式中,θa=θf-θsθfθs分别为脚和小腿与竖直方向夹角,∠BOF为设计的结构角度,设计值为84.75°.

  • 根据正弦定理和余弦定理可得:

  • BOA=arccosAO¯2+BO¯2-AB¯22AOBO¯COA=arcsinAC¯AO¯
    (5)
  • 联立式(1)~式(5),可得:

  • θa=π-arccosCO¯0+pθma2π2+AC¯2+BO¯2-AB2¯2CO¯0+pθma2π2+AC¯2BO¯-arcsinAC¯CO¯0+pθma2π2+AC¯2-BOF
    (6)
  • 式中,所设计的各连杆尺寸参数如下表1所示.

  • 表1 踝关节传动结构连杆设计值

  • Table1 Linkage design settings for the ankle joint transmission structure

  • 理论上,由式(6)可确定踝关节电机运动角度θma经过连杆结构后与踝关节相对运动θa角度的关系.但式(6)中存在反正弦和反余弦函数,无法直接求出显式表达式.我们用θma表示为θa的函数,并用隐函数表示为:

  • θma=fθa
    (7)
  • 式(7)两边同时对时间求导,可得:

  • θ˙ma=dfθadθaθ˙a
    (8)
  • 因此,根据传动比定义,踝关节传动比为:

  • ka=θ˙maθ˙a=dfθadθa
    (9)
  • 根据人体正常行走步态在矢状面上的踝关节屈曲角度[19]和已设计假肢[7]对踝关节角度要求,我们确定在θa=-π6π6范围内利用傅里叶级数拟合踝关节电机运动角度θma与踝关节相对运动角度θa的关系.图3分别展示了6、7、8阶傅里叶级数拟合结果,从图3(b)可以看出,8阶傅里叶级数的踝关节传动比拟合误差约为2.43×10-3,拟合已经足够精确.故拟合的傅里叶级数为:

  • θma=a0+i=18 aicosiωθa+bisiniωθa
    (10)
  • 式中,最终拟合的各系数如表2所示.

  • 式(10)两边对时间求导,得到传动比为:

  • ka=i=18 -iωaisiniωθa+iωbicosiωθa
    (11)
  • ka=gθa),拟合关系如下图3所示.由图3(a)可以看出,踝关节传动比与踝关节转角具有很强的非线性关系,因此在动力学参数辨识时,要考虑其对辨识结果的影响.

  • 表2 踝关节传动比的傅里叶级数拟合系数

  • Table2 Fourier series fitting coefficients for ankle joint transmission ratio

  • 图3 踝关节传动比拟合结果:(a)不同阶数傅里叶级数的拟合效果;(b)不同阶数傅里叶级数的拟合误差

  • Fig.3 Fitting results of ankle joint transmission ratio: (a) the fitting effect of different orders of Fourier series; (b) the fitting error of different orders of Fourier series

  • 2.2 基于拉格朗日方法的动力学建模

  • 按照文献[13]的辨识方式,把假肢膝关节上端固定,图1所示的动力假肢简化为图4所示的双摆杆结构.假设膝关节电机位于膝关节轴,质心位于关节原点上,踝关节电机位于小腿上,踝关节以下部分(下文叫做脚)质心位于踝关节竖直位置,同时忽略踝关节传动的拉杆和滚珠丝杠螺母的质量.

  • 图4 动力大腿假肢简化模型

  • Fig.4 Simplified model of the powered ankle-knee prosthesis

  • 定义小腿和脚连杆与竖直方向的绝对坐标为广义坐标q=(θsθfT,顺时针为正方向(从坐标轴指向连杆).利用Euler-Lagrange方程,建立图(4)所示大腿假肢在固定坐标系下的动力学方程为:

  • M(q)q¨+C(q,q˙)q˙+G(q)=T(q,q˙)
    (12)
  • 其中,质量阵MqR2×2各元素如下:

  • M1, 1=Imkkk2+Ik+Is+Im+msds2+mmdma2+mfls2+Imaka2, M1, 2=mflsdfcosθs-θf-Imaka2, M2, 1=mflsdfcosθs-θf-Imaka2, M2, 2=If+mfdf2+Imaka2.

  • 科氏力及离心力矩阵Cqq˙R2×2各元素为:

  • C1, 1=3Imakakaθaθ˙s-4Imakakaθaθ˙f, C1, 2=-2Imakakaθaθ˙s+3Imakakaθaθ˙f+mflsdfsinθs-θfθ˙f, C2, 1=-3Imakakaθaθ˙s+2Imakakaθaθ˙f-mflsdfsinθs-θfθ˙s, C2, 2=4Imakakaθaθ˙s-3Imakakaθaθ˙f.

  • 重力阵GqR2×1各元素如下:

  • G1, 1=msds+mmdma+mflsgsinθs, G2, 1=mfdfgsinθf.

  • 广义力阵Tqq˙R2×1,各元素如下:

  • T1, 1=Tms-Tτs-Tmf-Tτf, T2, 1=Tmf-Tτf.

  • 以上各式中,ls为小腿长度,dsdfdma分别为小腿质心到膝关节转动中心距离、脚质心到踝关节转动中心距离以及踝关节电机质心到膝关节转动中心的距离,msmf分别为小腿(包含电机定子)和脚的质量,mm为电机定子和转子总质量,mmkmma分别为膝关节和踝关节的电机转子及包含相连接转动轴质量,ImkIma分别为膝关节和踝关节的电机转子包含相连接转动轴的转动惯量,IsIf分别为小腿和脚相对于其自身质心的转动惯量,IkIa分别为膝关节和踝关节经减速器传动后的相关部件转动惯量,θmkθma分别为膝关节和踝关节电机的转动角度,kkka分别为膝关节和踝关节传动比,TmsTmf分别为膝关节和踝关节电机扭矩,TτsTτf分别为膝关节和踝关节的关节摩擦力损失.

  • 本文关节摩擦力采用常见的库伦-粘性摩擦力模型形式表征:

  • Tτs=fcssgnθ˙s+fvsθ˙s

  • Tτf=fcfsgnθ˙f+fvfθ˙f
    (13)
  • 式中,fcsfvsfcffvf分别为膝关节库伦粘性摩擦系数以及踝关节库伦和粘性摩擦系数.

  • 假肢动力学方程中共有17个参数,其中小腿长度ls采用游标卡尺精确测定为0.3615 m,小腿质量ms、脚的质量mf和电机定子转子的总质量mm采用电子秤进行精确测定,分别为1.624kg、0.442kg和0.4112kg.除了4个摩擦参数外的剩余9个参数可通过CAD软件测量估计其大小,以便跟辨识结果对比,测量结果如表3所示.

  • 根据参数的可辨识性,参数ImkIkIsIm可合并为一个待辨识参数,即I1=Imkkk2+Ik+Is+Im.因此,动力学方程中17个参数简化为包括4个摩擦参数在内的共10个待辨识参数,分别为dmadfdsI1ImaIffcsfcffvsfvf.

  • 表3 大腿假肢动力学参数CAD软件估计结果

  • Table3 CAD software estimations of the ankle-knee prosthesis dynamic parameters

  • 3 参数辨识

  • 3.1 辨识实验方案

  • 参数辨识实验如图5所示,膝关节上端通过固定杆悬挂于隔震台上,假肢上关节轴和连杆中心贴上不同颜色的标记点,利用相机拍摄记录假肢运动过程,拍摄帧率约为120Hz.同时假肢控制器记录电机上编码器数据、电机的电流数据和关节编码器数据,并通过无线串口发送给PC记录保存,记录频率约为200Hz.

  • 实验时膝关节电机和踝关节都采用位置控制模式,分别利用激励基频是2.5 Hz的5阶傅里叶级数激励信号测量3次,每次测量约2分钟,设置为:

  • i=15 Asin(5iπt)+Bcos(5iπt)
    (14)
  • 其中,幅值AB根据关节的机械限位进行调整.对于膝关节A取0.2,B取0.1,对于踝关节A取0.1,B取0.05.

  • 图5 动力大腿假肢参数辨识实验:(a)实验所使用器材;(b)实验时假肢的步态周期图

  • Fig.5 Parameter identification experiments: (a) equipment used in the experiment; (b) photos of gait cycles of the experimental prosthesis

  • 3.2 参数辨识方法

  • 本文参数辨识方法采用普遍使用的粒子群优化算法(PSO).相比于其它辨识算法,粒子群优化算法不需要求解梯度,算法更为简单,可以针对非光滑摩擦进行参数辨识.其还具有精度高、收敛速度快等优点,在参数辨识领域得到广泛应用.例如,Wang[20]将粒子群优化算法与不确定的速度和位移的系数修正相结合,构建了基于自适应信号修正的辨识方法.Chen[21]提出多段摩擦系数反演方法,并利用基于粒子群优化(PSO)的算法来计算了页岩气套管作业中的多段摩擦系数.本文的大腿假肢动力学参数辨识,也使用粒子群优化算法(PSO)进行参数辨识.首先通过位置激励假肢双关节获得运动信号,然后通过膝关节和踝关节驱动器获得关节电机电流数据,得到电机实测扭矩.电机扭矩计算为:

  • Tms=KimsTmf=Kimf
    (15)
  • 式中,K为电机扭矩常数,通过实验测量得到,取值为0.1031Nm/A.

  • 本文利用粒子群优化算法辨识的目标是寻找最优的一组动力学参数以最小化仿真所求关节扭矩与实测关节扭矩之间的误差,即:

  • minA1A10 1n Tmk,sim-Tmk, mea 2+Tma,sim-Tma , mea 2n
    (16)
  • 式中,A1···A10为待辨识参数,Tmk,simTma,sim为辨识仿真所求膝关节和踝关节电机扭矩,Tmk,meaTma,mea为实测膝关节和踝关节电机扭矩.

  • 3.3 实验数据处理

  • 首先利用MATLAB提取相机拍摄视频中每个标记点像素中心位置坐标.相机拍摄每帧视频原点为左上角,通过RGB值提取标记点的中点,依次提取四个标记点的像素坐标中心点位置.控制器记录的原始数据为电机上编码器数据、电机的电流数据和踝关节编码器数据.记录的编码器数据原始值为当前编码器刻度值,需要转换为弧度制.记录的电机电流数据需要乘以扭矩常数转换为电机扭矩.

  • 图6 参数辨识实验获得的数据:(a)关节角度;(b)关节角速度;(c)关节角加速度;(d)关节扭矩.黑色线表示膝关节,蓝色线表示踝关节

  • Fig.6 Data obtained from parameter identification experiments: (a) joint angle; (b) joint angular velocity; (c) joint angular acceleration; (d) joint torque. The black line indicates the knee joint and the blue line indicates the ankle joint

  • 由于实验只测得了膝关节和踝关节的角度信号,需要构造角速度和角加速度.由于多项式滑移窗口方法可以方便的重构出角度、角速度和角加速度信号,并且具有重构效果好,抗噪能力强等优点.故本文参考文献[2022]中的方法来获得辨识所需要的数据.为了拟合整个时程上多个周期的角度信号,通过调整Legendre多项式的阶数和拟合时间窗口的长度来提高拟合精度.本文设计的Legendre多项式拟合阶数为20,拟合窗口长度为1 s,相应的速度和加速度通过Legendre多项式求导得出.最终得到的小腿和脚连杆与竖直方向的绝对角度、绝对角速度、绝对角加速度和关节扭矩数据如图6所示.

  • 3.4 辨识结果

  • 最终得到的动力学参数辨识结果如表4所示.与表3的动力学参数CAD软件测量结果对比可知,除了踝关节电机到膝关节转轴距离辨识的偏差比较大之外,大部分参数比较接近.

  • 表4 大腿假肢动力学参数辨识结果

  • Table4 Dynamic parameter identification results for the ankle-knee prosthesis

  • 为了比较最终的辨识效果,我们分别把辨识后的动力学参数数据和CAD测量的动力学参数数据代入动力学方程,求出电机扭矩来对比验证辨识结果.如图7所示是假肢膝关节参数辨识结果的验证图,其中上图的黑线是实验测得的膝关节电机扭矩,蓝色线是CAD测量的动力学参数代入动力学方程求得的膝关节电机扭矩,红线是辨识后的动力学参数代入动力学方程求得的膝关节电机扭矩.从图7的下图中数据可以得出,由CAD测量的动力学参数计算出的电机扭矩与实测电机扭矩值之间的均方根误差RMSE为0.0076 Nm,而辨识后参数计算出的电机扭矩与实测电机扭矩值的均方根误差RMSE为0.00049 Nm,为辨识前膝关节电机扭矩均方根误差的6.45%,辨识后膝关节电机扭矩精度明显提高.

  • 如图8所示,其中上图的黑线是实验测得的踝关节电机扭矩,蓝色线是CAD测量的动力学参数代入动力学方程求得的踝关节电机扭矩,红线是辨识后的动力学参数代入动力学方程求得的踝关节电机扭矩.从图8的下图中数据可以得出,对于假肢踝关节,由CAD测量的动力学参数计算出的踝关节电机扭矩与实测踝关节电机扭矩的均方根误差RMSE为0.0012 Nm,而辨识后参数计算出的电机扭矩均方根误差RMSE为0.00023 Nm,为辨识前踝关节电机扭矩均方根误差的19.17%,辨识后的精度也有所提高,但精度提高并没有膝关节的多.总的来说,辨识的动力学参数结果比用CAD评估的参数更加准确,更能表示系统的动力学特性.

  • 图7 膝关节电机扭矩的实测及重构时程对比验证:(a)电机扭矩值;(b)电机扭矩误差.黑色线表示电机扭矩的实测值,蓝色线表示由CAD评估参数动力学求得的电机扭矩,红色线表示由辨识参数动力学求得的电机扭矩

  • Fig.7 Comparison and verification of the actual and reconstructed knee joint motor torques: (a) motor torque values; (b) motor torque error. The black line indicates the measured value of motor torque, the blue line indicates the motor torque derived from CAD evaluation of parametric dynamics, and the red line indicates the motor torque derived from discriminative parametric dynamics

  • 图8 踝关节电机扭矩的实测及重构时程对比验证:(a)电机扭矩值;(b)电机扭矩误差.黑色线表示电机扭矩的实测值,蓝色线表示由CAD评估参数动力学求得的电机扭矩,红色线表示由辨识参数动力学求得的电机扭矩

  • Fig.8 Comparison and verification of the actual and reconstructed knee joint motor torques: (a) motor torque values; (b) motor torque error. The black line indicates the measured value of motor torque, the blue line indicates the motor torque derived from CAD evaluation of parametric dynamics, and the red line indicates the motor torque derived from discriminative parametric dynamics

  • 4 结论

  • 本文主要研究实验室新设计的具有固定传动比膝关节和具有非线性传动比踝关节的大腿假肢动力学参数辨识问题.通过研究其动力学参数的CAD测量与实际辨识后参数所求电机扭矩的误差,明确了基于库伦-粘性摩擦的大腿假肢动力学参数的精度,进而为提高基于模型的假肢控制精度奠定了理论研究基础.通过本文的研究可以得出如下结论:

  • (1)通过对基于库伦-粘性摩擦模型的大腿假肢动力学进行参数辨识,可以很大程度上提高动力学预测电机扭矩的精度.

  • (2)对于具有不同传动链的不同动力假肢关节,多级传动链的累积误差会造成系统的不确定性放大,最终使动力学参数的辨识精度变差.

  • 参考文献

    • [1] 康艳梅.2010 年中国残疾人事业发展统计公报[R].北京:中国残疾人联合会,2010. KANG Y M,.2010 China disability development statistics bulletin [R].Beijing:China Disabled Persons' Federation,2010.(in Chinese)

    • [2] 王富军.中国糖尿病足防治指南(2019版)解读[J].河北医科大学学报,2019,40(11):1241-1245+1250.

    • [3] ZIEGLER-GRAHAM K,MACKENZIE E J,EPHRAIM P L,et al.Estimating the prevalence of limb loss in the United States:2005 to 2050 [J].Archives of Physical Medicine and Rehabilitation,2008,89(3):422-429.

    • [4] MILLER W C,DEATHE A B,SPEECHLEY M,et al.The influence of falling,fear of falling,and balance confidence on prosthetic mobility and social activity among individuals with a lower extremity amputation [J].Archives of Physical Medicine and Rehabilitation,2001,82(9):1238-1244.

    • [5] GABERT L,TRAN M,LENZI T.Design of an underactuated powered ankle and toe prosthesis [C]//2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society(EMBC).New York,USA:IEEE,2021:4920-4923.

    • [6] AZOCAR A F,MOONEY L M,DUVAL J F,et al.Design and clinical implementation of an open-source bionic leg [J].Nature Biomedical Engineering,2020,4:941-953.

    • [7] ELERY T,REZAZADEH S,NESLER C,et al.Design and validation of a powered knee-ankle prosthesis with high-torque,low-impedance actuators [J].IEEE Transactions on Robotics,2020,36(6):1649-1668.

    • [8] WANG X,ZHANG Y,LIANG W,et al.Design,control,and validation of a polycentric hybrid knee prosthesis [J].IEEE Transactions on Industrial Electronics,2023,70(9):9203-9214.

    • [9] ROUSE E J,VILLAGARAY-CARSKI N C,EMERSON R W,et al.Design and testing of a bionic dancing prosthesis [J].PLoS One,2015,10(8):e0135148.

    • [10] BHAKTA K,CAMARGO J,KUNAPULI P,et al.Impedance control strategies for enhancing sloped and level walking capabilities for individuals with transfemoral amputation using a powered multi-joint prosthesis [J].Military Medicine,2020,185:490-499.

    • [11] MA X,XU J,ZHANG X X.Bilateral constrained control for prosthesis walking on stochastically uneven terrain [J].International Journal of Mechanical Sciences,2023,239:107896.

    • [12] URREA C,PASCAL J.Design,simulation,comparison and evaluation of parameter identification methods for an industrial robot [J].Computers and Electrical Engineering,2018,67:791-806.

    • [13] LI Y B,GUAN X Y,LI W,et al.Dynamic parameter identification of a human-exoskeleton system with the motor torque data [J].IEEE Transactions on Medical Robotics and Bionics,2022,4(1):206-218.

    • [14] DALLALI H,FICANHA E,RASTGAAR AAGAAH M.Dynamic modeling of a 2-DOF cable driven powered ankle-foot prosthesis [C]//Proceedings of ASME 2016 Dynamic Systems and Control Conference.New York,USA:American Society of Mechanical Engineers,2017.

    • [15] YANG T,WU F,LIU M,et al.An optimization-based approach for prosthesis dynamic modeling and parameter identification [C]//Proceedings of ASME 2015 Dynamic Systems and Control Conference.New York,USA:American Society of Mechanical Engineers,2016.

    • [16] ISMAIL M A A,WINDELBERG J,LIU G J.Simplified sensorless torque estimation method for harmonic drive based electro-mechanical actuator [J].IEEE Robotics and Automation Letters,2021,6(2):835-840.

    • [17] BARTLETT H L,KING S T,GOLDFARB M,et al.Design and assist-as-needed control of a lightly powered prosthetic knee [J].IEEE Transactions on Medical Robotics and Bionics,2022,4(2):490-501.

    • [18] TRAN M,GABERT L,HOOD S,et al.A lightweight robotic leg prosthesis replicating the biomechanics of the knee,ankle,and toe joint [J].Science Robotics,2022,7(72):eabo3996.

    • [19] BROCKETT C L,CHAPMAN G J.Biomechanics of the ankle [J].Orthopaedics and Trauma,2016,30(3):232-238.

    • [20] WANG S Y,DIAO B B,ZHANG X X,et al.Adaptive signal-correction-based identification for friction perception of the vibration-driven limbless robot [J].Nonlinear Dynamics,2022,108(4):3817-3837.

    • [21] CHEN P J,GAO D L,WANG Z H,et al.Study on multi-segment friction factors inversion in extended-reach well based on an enhanced PSO model [J].Journal of Natural Gas Science and Engineering,2015,27:1780-1787.

    • [22] ZHANG X X,JI J C,FU J S,et al.Denoising identification for nonlinear systems with distorted streaming [J].International Journal of Non-Linear Mechanics,2019,117:103224.

  • 参考文献

    • [1] 康艳梅.2010 年中国残疾人事业发展统计公报[R].北京:中国残疾人联合会,2010. KANG Y M,.2010 China disability development statistics bulletin [R].Beijing:China Disabled Persons' Federation,2010.(in Chinese)

    • [2] 王富军.中国糖尿病足防治指南(2019版)解读[J].河北医科大学学报,2019,40(11):1241-1245+1250.

    • [3] ZIEGLER-GRAHAM K,MACKENZIE E J,EPHRAIM P L,et al.Estimating the prevalence of limb loss in the United States:2005 to 2050 [J].Archives of Physical Medicine and Rehabilitation,2008,89(3):422-429.

    • [4] MILLER W C,DEATHE A B,SPEECHLEY M,et al.The influence of falling,fear of falling,and balance confidence on prosthetic mobility and social activity among individuals with a lower extremity amputation [J].Archives of Physical Medicine and Rehabilitation,2001,82(9):1238-1244.

    • [5] GABERT L,TRAN M,LENZI T.Design of an underactuated powered ankle and toe prosthesis [C]//2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society(EMBC).New York,USA:IEEE,2021:4920-4923.

    • [6] AZOCAR A F,MOONEY L M,DUVAL J F,et al.Design and clinical implementation of an open-source bionic leg [J].Nature Biomedical Engineering,2020,4:941-953.

    • [7] ELERY T,REZAZADEH S,NESLER C,et al.Design and validation of a powered knee-ankle prosthesis with high-torque,low-impedance actuators [J].IEEE Transactions on Robotics,2020,36(6):1649-1668.

    • [8] WANG X,ZHANG Y,LIANG W,et al.Design,control,and validation of a polycentric hybrid knee prosthesis [J].IEEE Transactions on Industrial Electronics,2023,70(9):9203-9214.

    • [9] ROUSE E J,VILLAGARAY-CARSKI N C,EMERSON R W,et al.Design and testing of a bionic dancing prosthesis [J].PLoS One,2015,10(8):e0135148.

    • [10] BHAKTA K,CAMARGO J,KUNAPULI P,et al.Impedance control strategies for enhancing sloped and level walking capabilities for individuals with transfemoral amputation using a powered multi-joint prosthesis [J].Military Medicine,2020,185:490-499.

    • [11] MA X,XU J,ZHANG X X.Bilateral constrained control for prosthesis walking on stochastically uneven terrain [J].International Journal of Mechanical Sciences,2023,239:107896.

    • [12] URREA C,PASCAL J.Design,simulation,comparison and evaluation of parameter identification methods for an industrial robot [J].Computers and Electrical Engineering,2018,67:791-806.

    • [13] LI Y B,GUAN X Y,LI W,et al.Dynamic parameter identification of a human-exoskeleton system with the motor torque data [J].IEEE Transactions on Medical Robotics and Bionics,2022,4(1):206-218.

    • [14] DALLALI H,FICANHA E,RASTGAAR AAGAAH M.Dynamic modeling of a 2-DOF cable driven powered ankle-foot prosthesis [C]//Proceedings of ASME 2016 Dynamic Systems and Control Conference.New York,USA:American Society of Mechanical Engineers,2017.

    • [15] YANG T,WU F,LIU M,et al.An optimization-based approach for prosthesis dynamic modeling and parameter identification [C]//Proceedings of ASME 2015 Dynamic Systems and Control Conference.New York,USA:American Society of Mechanical Engineers,2016.

    • [16] ISMAIL M A A,WINDELBERG J,LIU G J.Simplified sensorless torque estimation method for harmonic drive based electro-mechanical actuator [J].IEEE Robotics and Automation Letters,2021,6(2):835-840.

    • [17] BARTLETT H L,KING S T,GOLDFARB M,et al.Design and assist-as-needed control of a lightly powered prosthetic knee [J].IEEE Transactions on Medical Robotics and Bionics,2022,4(2):490-501.

    • [18] TRAN M,GABERT L,HOOD S,et al.A lightweight robotic leg prosthesis replicating the biomechanics of the knee,ankle,and toe joint [J].Science Robotics,2022,7(72):eabo3996.

    • [19] BROCKETT C L,CHAPMAN G J.Biomechanics of the ankle [J].Orthopaedics and Trauma,2016,30(3):232-238.

    • [20] WANG S Y,DIAO B B,ZHANG X X,et al.Adaptive signal-correction-based identification for friction perception of the vibration-driven limbless robot [J].Nonlinear Dynamics,2022,108(4):3817-3837.

    • [21] CHEN P J,GAO D L,WANG Z H,et al.Study on multi-segment friction factors inversion in extended-reach well based on an enhanced PSO model [J].Journal of Natural Gas Science and Engineering,2015,27:1780-1787.

    • [22] ZHANG X X,JI J C,FU J S,et al.Denoising identification for nonlinear systems with distorted streaming [J].International Journal of Non-Linear Mechanics,2019,117:103224.

  • 微信公众号二维码

    手机版网站二维码