>> 自然科学版期刊 >> 2015年02期 >> 正文
基于小波包主元分析的表面肌电信号特征识别
供稿: 王红旗;李林伟;毛啊敏 时间: 2018-11-19 次数:

作者:王红旗李林伟;毛啊敏

作者单位:河南理工大学电气工程与自动化学院

摘要:针对表面肌电信号(SEMG)的非平稳性及小波包变换系数维数过高的问题,提出一种小波包主元分析和线性判别分析相结合的表面肌电信号动作特征识别新方法。以表面肌电信号用于智能轮椅为例,对采集到的两路SEMG信号进行小波包主元分析,提取SEMG信号的运动特征矩阵,并将运动特征矩阵输入到线性判别分类器进行分类,实现了前臂动作识别。试验表明:该方法能够将小波包系数矩阵由16维降到4维,并且对前臂的四种动作模式(握拳、展拳、手腕内翻和手腕外翻)的平均正确识别率达98%,与传统的小波包变换相比有较高的识别率。

基金:国家自然科学基金资助项目(61104119);河南理工大学博士基金资助(B2010-60);

关键词:表面肌电信号;小波包主元分析;线性判别分析;

DOI:10.16186/j.cnki.1673-9787.2015.02.020

分类号:R318.04;TN911.7

Abstract:Surface electromyography ( SEMG) is used as an example of intelligent wheelchair, in view of SEMG non-stationary characteristics and higher dimension of wavelet packet transform coefficients, put forward a new method that SEMG signal feature is through wavelet packet principal component analysis ( WPPCA) and linear discriminant analysis ( LDA) . Wavelet packet principal component analysis is employed to two SEMG signal and extract motion characteristic matrix of SEMG signal. Then motion characteristic matrix is put into linear discriminant classifier and forearm action recognition is realized. Experiments show that using this method can make the wavelet packet coefficient matrix of surface EMG signal consist of 16 dimensional to 4 dimensional, and successfully identify four kinds of motions such as hand grasping, hand opening, radial flexion and ulnar flexion, and the action recognition rate is up to 98%. Compared with the traditional wavelet packet transform, this algorithm has higher recognition rate.

最近更新