虚拟小波熵神经网络平台在SEMG分类中的应用

Application of virtual wavelet entropy NNT platform in sEMG classification

建立了以虚拟仪器为基础的表面肌电信号的分类平台。利用小波变换的时一频局部化信息特征及信息熵反映信源平均不确定性的特性,将小波分析和信息熵结合起来,并结合人工神经网络的自我学习能力,对表面肌电信号进行分类和识别。通过大量的表面肌电信号样本训练后,使网络不仅对训练样本有较高的分辨率(95%),而且对未训练过的样本也有较好的分辨率(90%)。训练结果表明,算法计算简单、实现容易。

A virtual instrument based on the classification of sEMG signal platform was established.By using the wavelet detection and information entropy,and combining the artificial neural network（NNT） as well as the self—learning capability of artificial neural networks,after training with a large number of ECG samples,the network not only has a high accuracy（95%）for the training samples,but also has a good accuracy（90%）for the not-trained samples.The proposed method is easy to comput,and suitable for the sEMG classification.