文献类型：期刊文献

英文题名：Study of Mechanomyography-Based Wrist Movement Classification with Repeatedly Wearing a Signal Acquisition Armband

作者：Zhang, Yue[1]; Zhao, Tongtong[1]; Zhang, Hanyang[1]; Xia, Chunming[1]; Cao, Gangsheng[1]

机构：[1] East China University of Science and Technology, Department of Mechanical Engineering, Shanghai, China

年份：2022

起止页码：771

外文期刊名：2022 IEEE Asia-Pacific Conference on Image Processing, Electronics and Computers, IPEC 2022

收录：EI(收录号：20222412229378)

语种：英文

外文关键词：Exoskeleton (Robotics) - Human robot interaction - Neural networks - Pattern recognition - Time domain analysis

摘要：For the field of rehabilitation engineering, pattern recognition of human movements based on mechanomyography (MMG) has certain application value in control of prosthetics, exoskeleton robot and human-machine interaction (HMI). Currently in most experiments, multi-channel MMG are collected by the method of just wearing signal acquisition device at one-time. However, repeatedly wearing the signal acquisition device will affect the signal since the sites of sensors and contact pressure inevitably changes. In this study, an experiment is designed to collect MMG of forearm by repeatedly wearing an acquisition band. support vector machine (SVM) and artificial neural network (ANN) are adopted to build the classifiers for wrist movements, and a novel cross validation method is utilized to evaluate the recognition results. When using 10 time domain features, the performance of ANN is approximate to SVM in the aspects of test subsets and subjects. And using 5 specific time domain feature subsets can obtain classification accuracy similar as using all 10 features. The results demonstrate that although the classification accuracy decreases, it is still feasible to achieve the recognition of wrist movement on the condition of repeatedly wearing the acquisition band. ? 2022 IEEE.