基于小波变换及AR模型的EMG模式识别研究

Wavelet transform and AR model based pattern recognition of EMG

当前表面肌电信号(SEMG)已经被广泛地应用到智能康复设备当中,然而在智能轮椅中的应用还处于起步阶段。为了实现SEMG在智能轮椅中的应用,提出了一种将小波变换和AR建模相结合的前额表面肌电信号分析方法。首先将采集得到的肌电信号进行预处理,获得有效数据段;然后,对信号进行小波分析,并将所得到的小波系数进行AR建模;最后,将得到的特征向量作为RBF神经网络的输入,对闭左眼、闭右眼、闭双眼和收下颚等4种动作进行运动模式分类。实验表明,这种信号分析方法兼具小波分析与AR建模的优点,取得了比较理想的识别效果。

Currently surface Electromyography signal（SEMG） has been widely applied in intelligent recovery equipments.But its application in Intelligent Wheelchair is just at the beginning.In order to realize SEMG＇s application in Intelligent Wheelchair a novel Wavelet Transform and AR model based SEMG analytical method has been proposed.First,the collected SEMG signals are preprocessed to get effective data segments.Then the signals are analysed by Wavelet Transform and AR modeling are carried on to the wavelet coefficients that are obtained.Finally,the feature vectors are took as inputs of a RBF Neural Network to classify four movements include closing left eye,closing right eye,closing both eyes and chawing.Experimental results indicate that this analytical method has both the advantages of Wavelet Transform and AR model,and can achieve ideal recognition effects.