基于表面肌电信号的上肢康复动作识别  

Upper Limb Rehabilitation Action Recognition Based on Surface EMG

作　　者：顾玉平 李宪华[2] 张康 罗耀 GU Yuping;LI Xianhua;ZHANG Kang;LUO Yao(School of Artificial Intelligence,Anhui University of Science and Technology,Huainan 232001,China;School of Mechanical and Electrical Engineering,Anhui University of Science and Technology,Huainan 232001,China)

机构地区：[1]安徽理工大学人工智能学院,安徽淮南232001 [2]安徽理工大学机电工程学院,安徽淮南232001

出　　处：《洛阳理工学院学报（自然科学版）》2024年第3期56-61,共6页Journal of Luoyang Institute of Science and Technology：Natural Science Edition

基　　金：安徽省重点研究与开发计划项目(2022i01020015).

摘　　要：选取上肢肱桡肌、尺侧弯曲肌、肱肌、肱二头肌、三角肌作为采集对象,使用干电极片以及数据采集卡采集表面肌电信号(sEMG)。将采集到的表面肌电信号进行预处理,提取时域特征、频域特征以及信息熵特征,将提取的特征用于卷积神经网络模型的训练,再抽取特征值部分样本作为验证集合进行交叉验证。实验结果表明,融合信息熵作为特征样本训练准确率高达91%,明显高于单一特征样本以及融合时频域特征样本。In this paper,the brachioradialis muscle,ulnar flexor muscle,brachialis muscle,biceps brachialis muscle and deltoid muscle of upper limb were selected as the collection objects,and the surface EMG signal was collected by dry electrode and data acquisition card.The time domain features,frequency domain features and information entropy features are extracted after pre-processing the collected surface EMG signals,and the extracted features are used to train the convolutional neural network model.Then some samples of eigenvalues are selected as verification sets for cross-verification.The experimental results show that the training accuracy of fusion information entropy as a feature sample is as high as 91%,which is obviously higher than that of single feature sample and fusion time-frequency domain feature sample.

关 键 词：表面肌电信号 模式识别 康复运动 卷积神经网络 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]