基于SEMG的膝关节康复动作模式识别技术研究

摘要

This paper determines to use RMS,fourth-order AR model parameter for SEMG feature extraction method and proposes the formation of a new fifth-order feature vectors. After designing lower limb movement pattern and extracting eigenvalues of EMG,BP neural network classifier is designed and the three extracted feature vectors are used as the classifier input. Experimental result shows that BP neural network can be used to accurately identify the knee action,when the new input vector is entered in the classifi￣er,its training efficiency is high.%确定RMS、四阶AR模型参数作为表面肌电信号的特征提取方法，提出组建五阶新特征向量。设计下肢动作模式实验，提取肌电信号的特征值，设计BP神经网络分类器，并将所提取的三种特征向量作为分类器的输入。实验表明BP神经网络可以准确识别膝关节动作，且新向量作为分类器输入时，训练效率高。