基于sEMG和GRNN的手部输出力估计

发布时间：2018-04-02 03:04

本文选题：表面肌电信号　切入点：广义回归神经网络　出处：《仪器仪表学报》2017年01期

【摘要】：针对智能肌电假手力控制的需要,提出一种基于表面肌电信号(s EMG)和广义回归神经网络(GRNN)的手部输出力估计方法。首先在介绍实验平台的基础上详细描述了肌电信号的采集和特征提取方法以及广义回归神经网络的构建;然后,通过在手臂8个不同部位粘贴肌电传感器来检测手部动作过程中的肌电信号;同时为了全面测量人手在三维空间中的输出力,采用三维力传感器对手部的输出力进行测量;在同步获得手臂上的多通道肌电信号(X)和手部三维力推拉信号(F)后,对采集得到肌电信号进行了特征提取得到特征矩阵X_F;将X_F和F用于构建GRNN网络,并用均方差和残差绝对值均值对手部输出力的估计结果进行评估。为验证该方法的有效性,进行了实验验证,结果表明,该方法能够很好地利用sEMG对手部的输出力进行估计。
[Abstract]:Aiming at the need of force control of intelligent myoelectric prosthetic hand, A method of hand output force estimation based on surface electromyography (EMG) and generalized regression neural network (GRNN) is proposed. Firstly, the acquisition and feature extraction method of EMG signal and its wide range are described in detail on the basis of the experimental platform. The construction of semantic regression neural network; Then, EMG signals are detected during hand movements by attaching EMG sensors to eight different parts of the arm, and in order to fully measure the output force of the hand in three-dimensional space, The output force of the hand was measured by a three dimensional force sensor, and after synchronous acquisition of the multichannel electromyography (EMG) signal on the arm and the hand force push pull signal (F), In order to verify the effectiveness of the proposed method, the feature matrix XFs is extracted from the acquired EMG signals, and the GRNN network is constructed by using XIF and F, and the estimation results of the output force of the hand are evaluated with the mean mean of mean value of mean square error (RMS) and residuals. The experimental results show that the method can estimate the output force of the hand by using sEMG.
【作者单位】：东南大学仪器科学与工程学院;南京航空航天大学自动化学院;
【基金】：国家自然科学基金(61325018,61663027) 江苏省科技支撑计划(BE2014132)项目资助
【分类号】：R318.17;TP183

【相似文献】