摘要
目的为实现不同上肢损伤程度患者的康复训练,设计一种基于表面肌电信号(surface electromyographic signal,sEMG)的上肢外骨骼康复训练系统。方法该康复训练系统的机械结构主要由背部控制部分、可变刚度驱动器以及可调节上肢支架部分组成;控制系统包括肌电采集、滤波、特征提取和动作分类识别。首先采集肌电信号,并提取时域特征;然后采用主成分分析法(principal component analysis,PCA)进行降维处理,用K均值聚类算法(K-means clustering algorithm)进行动作模式分类识别;最后对可变刚度驱动器进行刚度测量实验,并进行仿真实验验证分类效果。结果康复训练系统可以自主地调节刚度,并且动作模式的总体识别率为8974%。结论该康复训练系统动作模式识别率高,可以更好地带动患者完成康复训练。
Objective In order to realize the rehabilitation training of patients with different degrees of upper limb injury,an upper limb exoskeleton rehabilitation training system based on surface electromyographic signal(sEMG)is designed.Methods The mechanical structure of the rehabilitation training system is mainly composed of the back control part,the variable stiffness driver and the adjustable upper limb support part.The control system includes EMG acquisition,filtering,feature extraction and action classification and recognition.We first collect electromyographic signals and extract their time-domain features;Then,principal component analysis(PCA)is used for dimensionality reduction,and K-means clustering algorithm is used for action pattern classification and recognition;Finally,the stiffness measurement experiment of variable stiffness actuator is carried out,and the simulation experiment is carried out to verify the classification effect.Results The rehabilitation training system can adjust the stiffness independently,and the overall recognition rate of action pattern is 8974%.Conclusions The rehabilitation training system has a high recognition rate of movement patterns,which can better drive patients to complete rehabilitation training.
作者
付强
张志辉
张松源
段佰龙
FU Qiang;ZHANG Zhihui;ZHANG Songyuan;DUAN Bailong(International Joint Research Center,Tianjin University of Technology,Tianjin 300380;State Key Laboratory of Robotics and System,Harbin Institute of Technology,Harbin 150001;The Second Hospital of Harbin,Harbin 150056)
出处
《北京生物医学工程》
2024年第1期29-34,共6页
Beijing Biomedical Engineering
基金
天津市科技计划项目(18PTZWHZ00090)
天津市复杂系统控制理论与应用重点实验室项目(TJKL⁃CTACS⁃201903)资助。
关键词
表面肌电信号
康复训练
变刚度
特征提取
分类识别
surface EMG
rehabilitation training
variable stiffness
feature extraction
classification recognition
