Please wait a minute...
浙江大学学报(工学版)
浙江大学学报(工学版)  2021, Vol. 55 Issue (1): 89-95    DOI: 10.3785/j.issn.1008-973X.2021.01.011
计算机技术、自动控制技术     
基于相位划分的下肢连续运动预测
段有康1,2(),陈小刚1,*(),桂剑3,马斌3,4,李顺芬1,宋志棠1
1. 中国科学院 上海微系统与信息技术研究所，上海 200050
2. 中国科学院大学，北京 100049
3. 中国科学院 上海高等研究院，上海 200125
4. 上海中研久弋科技有限公司，上海 201200
Continuous kinematics prediction of lower limbs based on phase division
You-kang DUAN1,2(),Xiao-gang CHEN1,*(),Jian GUI3,Bin MA3,4,Shun-fen LI1,Zhi-tang SONG1
1. Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, Shanghai 200050, China
2. University of Chinese Academy of Sciences, Beijing 100049, China
3. Shanghai Advanced Research Institute, Chinese Academy of Sciences, Shanghai 200125, China
4. China Research Institute JIU YI, Shanghai 201200, China
 全文: PDF(1561 KB)   HTML
摘要：

为了实现针对特定个体的运动特点进行更精确的下肢连续运动预测和用更短的时间开展预测模型训练，采用对每个步态相位都建立预测模型的方法. 在识别出当前的步态相位后，使用当前相位的预测模型进行关节角度的预测. 使用支持向量机（SVM），对提出的方法进行验证. 实验表明，采用基于相位划分的下肢连续运动预测方法相比于对整个运动状态进行关节角度建模的预测方法，具有更高的预测精度和更短的模型训练时间. 髋、膝、踝关节的预测结果与真实值的相关系数均大于0.99，每次预测的角度与真实值的平均均方根误差均小于2°，训练时间缩短4.0~5.0倍.
关键词： 相位划分;  运动意图识别;  关节角度预测;  外骨骼机器人;  表面肌电信号    
Abstract:

A method of establishing a prediction model for each gait phase was adopted in order to achieve more accurate prediction of continuous movement of lower limbs for specific individual movement characteristics and use a shorter time to train the prediction model. The prediction model of the current phase was used to predict the joint angle after identifying the current gait phase. Support vector machine (SVM) was used to verify the proposed method. The experimental results show that the prediction method of continuous motion of lower limbs based on phase division has higher prediction accuracy and shorter model training time than the prediction method of joint angle modeling of the entire motion state. The correlation coefficients between the predicted results of the hip, knee, and ankle joints and the true values are all greater than 0.99. The average root mean square error of the predicted angle values and the true values is less than 2° each time, and the training time is shortened by 4.0~5.0 times.
Key words: phase division    motion intention recognition    joint angle prediction    exoskeleton robot    surface electromyography
收稿日期: 2020-05-12 出版日期: 2021-01-05
CLC:  V 445  
基金资助: 国家重点研发计划资助项目（2017YFA0206104）；军民融合专项资助项目（2019JMRH1KJ22）
通讯作者: 陈小刚     E-mail: dyk@mail.sim.ac.cn;chenxg@mail.sim.ac.cn
作者简介: 段有康（1994—），男，硕士生，从事新型存储器应用和机器学习研究. orcid.org/0000-0001-5379-7234. E-mail： dyk@mail.sim.ac.cn
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章　　
段有康
陈小刚
桂剑
马斌
李顺芬
宋志棠

引用本文:

段有康,陈小刚,桂剑,马斌,李顺芬,宋志棠. 基于相位划分的下肢连续运动预测[J]. 浙江大学学报(工学版), 2021, 55(1): 89-95.

You-kang DUAN,Xiao-gang CHEN,Jian GUI,Bin MA,Shun-fen LI,Zhi-tang SONG. Continuous kinematics prediction of lower limbs based on phase division. Journal of ZheJiang University (Engineering Science), 2021, 55(1): 89-95.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2021.01.011    或    http://www.zjujournals.com/eng/CN/Y2021/V55/I1/89

图 1  几种运动步态周期划分示意图
图 2  基于相位划分的下肢连续运动预测算法流程图
图 3  腓肠肌内侧几种运动状态肌电信号图
图 4  上坡和慢跑实验
图 5  步态相位识别
图 6  整段数据进行预测建模的速度为6 km/h慢跑状态关节角度预测
图 7  基于相位划分的速度为6 km/h慢跑状态关节角度预测
方法 C $ \overline R $
本文方法 0.99 1°
文献[7]方法 0.93 5.83°
文献[10]方法 0.92 6.87°
文献[17]方法 0.96 2.37°
表 1  本文方法与其他方法预测精度的比较
Ntr $t/{\rm{s}}$ $T/{\rm{s}}$ A
500 0.1323 0.0358 3.70
1000 0.4130 0.0835 4.95
1500 0.8443 0.1587 5.32
2000 1.2419 0.2762 4.50
2500 2.1432 0.4579 4.68
3000 2.8276 0.6230 4.54
3500 3.6642 0.8696 4.21
4000 4.8305 1.1120 4.34
表 2  2种算法的训练时间比较
1 龙亿. 下肢外骨骼人体运动预测与人机协调控制技术研究[D]. 哈尔滨: 哈尔滨工业大学, 2017.
LONG Yi. Research on human body motion prediction and human-machine coordinated control technology of lower exoskeletonx [D]. Harbin: Harbin Institute of Technology, 2017.
2 俞毕洪. 基于表面肌电信号的穿戴式人体行为识别系统[D]. 杭州: 浙江大学, 2018.
YU Bi-hong. Wearable human behavior recognition system based on surface EMG signals [D]. Hangzhou: Zhejiang University, 2018.
3 丁其川, 熊安斌, 赵新刚, 等 基于表面肌电的运动意图识别方法研究及应用综述[J]. 自动化学报, 2016, 42 (1): 13- 25
DING Qi-chuan, XIONG An-bin, ZHAO Xin-gang, et al Research and application review of motion intention recognition method based on surface EMG[J]. Journal of Automation, 2016, 42 (1): 13- 25
4 AI Q, ZHANG Y, QI W, et al Research on lower limb motion recognition based on fusion of sEMG and accelerometer signals[J]. Symmetry, 2017, 9 (8): 147- 166
doi: 10.3390/sym9080147
5 ZHANG F, LI P F, HOU Z G, et al sEMG-based continuous estimation of joint angles of human legs by using BP neural network[J]. Neurocomputing, 2011, 78 (1): 139- 148
6 隋修武, 石峰 基于改进MPSO-SVM算法的下肢连续运动预测模型[J]. 天津工业大学学报, 2019, 38 (6): 69- 75
SUI Xiu-wu, SHI Feng Prediction model of lower limb continuous motion based on improved MPSO-SVM algorithm[J]. Journal of Tianjin University of Technology, 2019, 38 (6): 69- 75
doi: 10.3969/j.issn.1671-024x.2019.06.012
7 徐俊武, 沈林勇, 章亚男, 等 利用健肢sEMG信号对康复机械腿进行映射控制[J]. 工业控制计算机, 2019, 32 (5): 8- 10
XU Jun-wu, SHEN Lin-yong, ZHANG Ya-nan, et al Mapping control of rehabilitation mechanical legs using sEMG signals of healthy limbs[J]. Industrial Control Computer, 2019, 32 (5): 8- 10
doi: 10.3969/j.issn.1001-182X.2019.05.004
8 TIPPING M E Sparse bayesian learning and the relevance vector machine[J]. Journal of Machine Learning Research, 2001, 1 (3): 211- 244
9 AU S K, WEBER J, HERR H Powered ankle foot prosthesis improves walking metabolic economy[J]. IEEE Transactions on Robotics, 2009, 25 (1): 51?66
10 张琴, 骆无意, 黄波, 等 基于肌电信号的下肢关节连续运动预测[J]. 华中科技大学学报: 自然科学版, 2017, 45 (10): 128- 132
ZHANG Qin, LUO Wu-yi, HUANG Bo, et al Prediction of continuous motion of lower limb joints based on EMG signals[J]. Journal of Huazhong University of Science and Technology: Natural Science Edition, 2017, 45 (10): 128- 132
11 CAI S, CHEN Y, HUANG S, et al SVM-based classification of sEMG signals for upper-limb self-rehabilitation training[J]. Front Neurorobot, 2019, 13: 31
doi: 10.3389/fnbot.2019.00031
12 陈树, 张继中 基于自适应均值粒子群算法的 SVM 参数优化方法[J]. 测控技术, 2018, 37 (4): 6- 10
CHEN Shu, ZHANG Ji-zhong SVM parameter optimization method based on adaptive mean particle swarm optimization[J]. Measurement and Control Technology, 2018, 37 (4): 6- 10
doi: 10.3969/j.issn.1000-8829.2018.04.002
13 范光辉, 葛科铎, 王璐, 等 基于下肢表面肌电信号的动作模式识别研究[J]. 齐齐哈尔大学学报: 自然科学版, 2018, 34 (6): 44- 48
FAN Guang-hui, GE Ke-duo, WANG Lu, et al Research on the action pattern recognition based on the electromyographic signal of the lower limb surface[J]. Journal of Qiqihar University: Natural Science Edition, 2018, 34 (6): 44- 48
14 SHULTZ A H, LAWSON B E, GOLDFARB M Variable cadence walking and ground adaptive standing with a powered ankle prosthesis[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2016, 24 (4): 495- 505
doi: 10.1109/TNSRE.2015.2428196
15 张仁志, 崔慧娟 基于短时能量的语音端点检测算法研究[J]. 电声技术, 2005, 7: 52- 54
ZHANG Ren-zhi, CUI Hui-juan Research on speech endpoint detection algorithm based on short-term energy[J]. Electronic Acoustic Technology, 2005, 7: 52- 54
doi: 10.3969/j.issn.1002-8684.2005.07.015
16 卢祖能, 曾庆杏, 李承晏, 等. 实用肌电图学[M]. 北京: 人民卫生出版社, 2000: 424-434.
17 戴虹, 钱晋武, 张震, 等 GRNN在肌电预测踝关节运动中的应用[J]. 仪器仪表学报, 2013, 34 (4): 845- 852
DAI Hong, QIAN Jin-wu, ZHANG Zhen, et al Application of GRNN in ankle movement prediction based on surface electromyography[J]. Chinese Journal of Scientific Instrument, 2013, 34 (4): 845- 852
doi: 10.3969/j.issn.0254-3087.2013.04.020
[1] 都明宇, 鲍官军, 杨庆华, 王志恒, 张立彬. 基于改进支持向量机的人手动作模式识别方法[J]. 浙江大学学报(工学版), 2018, 52(7): 1239-1246.
[2] 赵翠莲, 徐浩宇, 罗林辉, 王凯. 熵在不同等级偏瘫患者sEMG运动检测中的应用[J]. 浙江大学学报(工学版), 2018, 52(4): 798-805.
[3] 刘磊, 杨鹏, 刘作军. 采用多核相关向量机的人体步态识别[J]. 浙江大学学报(工学版), 2017, 51(3): 562-571.
[4] 佘青山, 孟明, 罗志增, 马玉良. 基于多核学习的下肢肌电信号动作识别[J]. J4, 2010, 44(7): 1292-1297.
[5] 张海红 王健 杨红春. 肱二头肌离心和向心收缩的肌电信号特征[J]. J4, 2006, 40(11): 1937-1941.
[6] 张海红 王健 杨红春. 肱二头肌离心和向心收缩的肌电信号特征[J]. J4, 2006, 40(11): 1937-1941.