基于新型双输入卷积神经网络的肌电模式识别

sEMG pattern recognition based on a novel dual-input convolutional neural network

针对双臂协同运动中蕴含的运动信息量大,难以充分解读且识别率不高的问题,提出一种新型的双输入卷积神经网络(ND-CNN)模型。首先,根据双臂运动的特点,分别设计数据整理和模型输入两种策略。然后,利用两个结构相同、参数共享的特征提取层提取信号本身的特征和信号之间的差别特征。最后,利用所提取的两类特征实现双臂协同动作的识别。在自主设计的双臂实验中,将ND-CNN与其余3种先进的神经网络对比。实验结果表明,本文所提的ND-CNN模型在识别精度和可靠性上优于其他网络模型,能够对双臂肌电动作有效识别。

A novel dual-input convolutional neural network(ND-CNN) model is proposed to solve the problems of large amount of motion information contained in dual-arm cooperative motion, difficult to fully interpret and low recognition rate. According to the characteristics of dual-arm motion, two strategies of data sorting and model input are designed, and then, two feature extraction layers with the same structure and shared parameters are used to extract the features of the signal itself and the discriminative features between the signals. Finally, the two kinds of extracted features are applied to realize the recognition of dual-arm cooperative action. In the self-designed dual-arm experiment, ND-CNN is compared with the other 3 advanced neural networks. The experimental results show that the proposed ND-CNN model is superior to the other network models in recognition accuracy and reliability, and that it can effectively recognize the dual-arm surface electromyography signals.