基于长短期记忆模型的外骨骼实时步态分类  

Real time gait phase classification of exoskeleton based on a long-short term memory model

作　　者：郭子铭 宋春宁[1] 王灿[2] GUO Zi-ming;SONG Chun-ning;WANG Can(College of Electrical Engineering, Guangxi University, Nanning 530004, China;Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, China)

机构地区：[1]广西大学电气工程学院,广西南宁530004 [2]中国科学院深圳先进技术研究院,广东深圳518055

出　　处：《广西大学学报（自然科学版）》2020年第5期1171-1179,共9页Journal of Guangxi University（Natural Science Edition）

基　　金：国家自然科学基金资助项目(51767005,2017YBF1302303);广西自然科学基金资助项目(2016GXNSFAA380328)。

摘　　要：外骨骼机器人作为一种创新的辅助技术正在蓬勃发展,它可以帮助偏瘫患者康复以再次行走。针对外骨骼机器人,提出了一种基于表面肌电图(sEMG)信号的长短期记忆模型(LSTMM)来识别4种不同的步态阶段。其中包括脚跟着地(HS)、脚掌着地(FF)、脚尖着地(HO)以及脚部抬起(SW),该模型仅需要6个sEMG信号。本研究使用Vicon-nexus(VN)验证了sEMG数据采集系统的可用性,避免了复杂的传感器系统,并确保了步态相位分类的准确性。最后,结果表明:当使用6个sEMG信号时,用于步态相位分类的LSTMM的精度为91.44%,优于其他算法。在用于不同人群时,对于步态阶段分类具有明显更高的预测精度和更好的鲁棒性。Exoskeleton robots are gaining momentum as an innovative assistive technology to help rehabilitate hemiplegia patients to walk again.A long-short term memory model(LSTMM)is proposed for an exoskeleton robot based on surface electromyography(sEMG)signals to recognize four different kinds of gait phases,including heel strike(HS),foot flat(FF),heel off(HO),and the swing(SW)between the foot and the ground for an exoskeleton robot.This model requires six sEMG signals to help hemiplegia patients walk.This study uses Vicon-nexus(VN)to verify the availability of sEMG data acquisition systems,avoid complex sensor systems,and ensure the accuracy of gait phase classification.Finally,the results show that the accuracy of the LSTMM used for gait phase classification is 91.44%when six sEMG signals are used,which is better than other algorithms.When used in different populations,it has significantly higher prediction accuracy and better robustness for gait phase classification.

关 键 词：长期短期记忆 步态分类 SEMG信号 鲁棒性 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]