摘要
基于表面肌电信号的肢体运动模式识别是假手仿生控制的基础,SEMG的个体差异与识别率是肌电假手实用化必须面对的问题。本文根据SEMG的频谱特性提出了一种新的特征提取方法———功率谱比值法。该方法的主要特点是以实时取得的SEMG功率谱信号为基础,确定最大功率谱附近的谱能量与全信号段谱能量之比为特征值,将人的个体差异影响降低到最低程度。模式分类器采用特别设计的Bayes统计决策算法,该方法在非特定人的条件下应用于前臂肌群的多运动模式识别时,识别正确率达到84%,已具备一定的实用性。
Movement recognition based on surface electromyography is the foundation of myoelectric prosthesis control. Individual difference and pattern identification rate of SEMG are the two problems that researchers must face to. Based on the spectrum characteristic of SEMG, a new feature extraction method called Power Spectrum Ratio Method was proposed. The method defines the ratio of maximum energy band spectrum to total power spectrum as an eigenvalue of SEMG, which decreases the individual difference effect to the lowest. A mode classifier was used for hand movement pattern recognition, which adopts the special designed Bayes statistics algorithm. When used in hand movement pattern recognition for no-specific person, this method reached a correctness of 84%, which is rather practical.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2006年第9期996-999,共4页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(60474054)
新世纪优秀人才支持计划(NCET-04-0558)资助项目
关键词
表面肌电信号
功率谱
Bayes统计决策算法
surface electromyography power spectrum Bayes statistics decision making algorithm
