基于Elman神经网络的肌电信号分类

The Classification of on the Surface EMG with Elman Network

BP网络有易陷入局部极小值的缺点。文章采用Elman人工神经网络,结合小波理论,提取了肌电信号的特征,对四组动作进行了识别。实验结果表明,此种方法显示了很好的分类效果,提高了收敛速度。

The drawback of BP network lies in its local minimum. The character of the surface EMG was detected and at the same time four different kinds of actions were picked up with Elman RNN methods and wavelet analysis. The experiment result is inspiring and its accuracy of prediction by Elman is higher than BP network.