| 基于有监督Kohonen神经网络的步态识别 |
| |
| 引用本文： | 郭欣,王蕾,宣伯凯,李彩萍.基于有监督Kohonen神经网络的步态识别[J].自动化学报,2017,43(3):430-438. |
| |
| 作者姓名： | 郭欣 王蕾 宣伯凯 李彩萍 |
| |
| 作者单位： | 1.河北工业大学控制科学与工程学院 天津 300130 |
| |
| 基金项目： | 河北省青年自然基金（F2016202327），河北省高等学校科学技术研究项目（Q2012079，ZC2016020），中国科学院人机智能协同系统重点实验室开放基金资助 |
| |
| 摘 要： | 表面肌电信号随着时间的变化而改变，这将影响运动模式的分类精度.传统人体下肢假肢运动模式的识别算法不能保证在整个肌电控制时间内达到对运动模式的有效识别.为了解决这些问题，本文提取步态初期200ms的信号的特征值，将无监督和有监督的Kohonen神经网络算法应用到大腿截肢者残肢侧的步态识别中，并与传统BP神经网络进行了对比.结果表明，有监督的Kohonen神经网络算法将五种路况下步态的平均识别率提高到88.4%，优于无监督的Kohonen神经网络算法和BP神经网络.
|
| 关 键 词： | 表面肌电信号 智能假肢 特征提取 有监督Kohonen神经网络 步态识别 |
| 收稿时间： | 2016-02-04 |
Gait Recognition Based on Supervised Kohonen Neural Network |
| |
| GUO Xin,WANG Lei,XUAN Bo-Kai,LI Cai-Ping.Gait Recognition Based on Supervised Kohonen Neural Network[J].Acta Automatica Sinica,2017,43(3):430-438. |
| |
| Authors: | GUO Xin WANG Lei XUAN Bo-Kai LI Cai-Ping |
| |
| Affiliation: | 1.School of Control Science and Engineering, Hebei University of Technology, Tianjin 3001302.Engineering Research Center of Intelligent Rehabilitation, Ministry of Education, Tianjin 300130 |
| |
| Abstract: | Surface electromyography (sEMG) is changeable with time, which will affect the classification accuracy. The traditional recognition method cannot guarantee its effectiveness within whole control cycle for lower limb movement. This paper extracts the feature from initial 200ms EMG, applies Kohonen and supervised Kohonen neural networks, and compares the result with BP neural network. Experimental results show that supervised Kohonen neural network is superior to the other two algorithms. The average recognition rate can be increased to 88.4% for five kinds of terrains. |
| |
| Keywords: | Surface electromyography (sEMG) intelligent prosthesis feature extraction supervised Kohonen neural network gait recognition |
|
| 点击此处可从《自动化学报》浏览原始摘要信息 |
|
点击此处可从《自动化学报》下载全文 |
