基于sEMG与足底压力信号融合的跌倒检测研究  

Study on fall detection based on surface EMG and plantar pressure signal fusion

作　　者：席旭刚[1] 武昊[1] 左静[1] 罗志增[1] 

机构地区：[1]杭州电子科技大学智能控制与机器人研究所,杭州310018

出　　处：《仪器仪表学报》2015年第9期2044-2049,共6页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(60903084;61172134);浙江省自然科学基金(LY13F030017);浙江省科技计划(2014C33105)项目资助

摘　　要：跌倒已经成为一种普遍危害老年人身心健康的事故,需要得到及时救治。设计了一种基于表面肌电(s EMG)和足底压力信号融合的跌倒检测系统。提取s EMG的近似熵及基本尺度熵特征,并根据足底压力的变化规律,提取动作信号段的压力特征,通过D-S证据推理将肌电信号与足底压力信号的SVM决策融合获得综合判别结果。实验结果表明,该方法对跌倒与ADL的平均识别率达到了91.7%,优于单一信源识别结果。Fall accidents have hecome a common physical and mental heahh hazard for the ehterly, and need to receive timely medical treatment. This paper presents a fall detection system based on sEMG and plantar pressure signal fusion. The sEMG apprnximate entropy and basic scale entropy are extracted. The pressure feature of action signal segment is extracted according to the rules of the plantar pressure change. The SVM decision fusion of the sEMG and plantar pressure signal is conducted through DS evidential reasoning to obtain the comprehensive recognition results. The experiment results show that the average recognition rates of fall and ADL reach 91.7% , which is higher than the recognition results of single source recognition.

关 键 词：跌倒检测 表面肌电 足底压力 支持向量机 信息融合 

分 类 号：TP212.3[自动化与计算机技术—检测技术与自动化装置] TH7[自动化与计算机技术—控制科学与工程]