摘要
设计了基于表面肌电信号的智能轮椅人机交互系统,首先通过CyberLink肌电传感器,对面部运动信号进行采集与分析处理,采用了深度信任网络(deep belief network,DBN)算法对肌电信号进行分类,进而用于智能轮椅的运动控制.实验表明:与支持向量机相比,用深度信任网络训练肌电信号,能有效地处理大量的肌电样本信号,并得到最高可达95.25%的识别率,提高了肌电信号的识别率、有效降低了对大量数据的处理时间、增强了智能轮椅响应的实时性.
On the basis of intelligent wheelchair controlled by signals from sEMG(surface electromyography).Firstly,sEMG signals generated by the facial movements were collected and analyzed by using a CyberLink device.DBN(deep belief network)was used to recognize different movement patterns of sEMG signal.An information accessibility HMI was designed to map facial movement patterns into corresponding control commands.Experiments show that compared to SVM(support vector machine),DBN can effectively deal with a large number of sEMG signal samples,and the highest recognition rate can reach 95.25%,which effectively reduces the processing time of a large amount of data,enhances the real-time of the intelligent wheelchair,and improves recognition rate of the sEMG signal.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2015年第S1期74-77,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(60905066
51075420)
科技部国际合作项目(2010DFA12160)
