刘晓光,李奂良,娄存广,梁 铁,王立玲,刘秀玲,王洪瑞.基于支持向量机的表面肌电信号和加速度融合跌倒识别方法[J].,2020,(2):385-390 |
基于支持向量机的表面肌电信号和加速度融合跌倒识别方法 |
Fall Recognition Based on Surface Electromyography and Acceleration Fusion by Support Vector Machine |
投稿时间:2019-02-06 修订日期:2019-02-28 |
DOI:10.13241/j.cnki.pmb.2020.02.039 |
中文关键词: 跌倒识别 表面肌电信号 三轴加速度 支持向量机 模式识别 |
英文关键词: Fall recognition Surface electromyography signal Triaxial acceleration Support vector machine Pattern recognition |
基金项目:国家自然科学基金项目(61473112;61673158);国家重点研发计划项目(2017YFB1401200);河北省博士后基金(B2019005001) |
|
摘要点击次数: 708 |
全文下载次数: 720 |
中文摘要: |
摘要 目的:针对老人易跌倒和跌倒过后可能产生严重后果这一现实问题,通过将表面肌电信号和加速度融合,进一步优化采用支持向量机分类器下的包含跌倒在内的几种不同动作的分类效果。方法:提出基于表面肌电和加速度信号融合的跌倒识别算法,首先采集股直肌,股内侧肌,胫骨前肌和腓肠肌的表面肌电信号以及位于腰部的三轴加速度信号作为实验数据,然后利用滑动窗口法提取表面肌电和加速度信号的均方根值,最后针对人体日常活动和跌倒的运动特征,构建了支持向量机的分类器。结果:实验数据共计320组数据,包括3种日常活动和向前跌倒,其中160组数据作为训练集,另外160组数据作为测试集。对4种动作进行识别实验,算法的准确度为93.23%、灵敏度为92.4%、特异度为100%,达到了良好的分类效果。结论:基于支持向量机的表面肌电信号和加速度融合的跌倒识别算法分类效果良好,对于老人跌倒防护具有现实意义。 |
英文摘要: |
ABSTRACT Objective: Aiming at the fact that the elderly are prone to fall and may have serious consequences after falling, through the fusion of surface electromyography signal and acceleration, the classification effect of several different actions including falling under the support vector machine classifier was further optimized. Methods: A fall recognition algorithm based on surface electromyography and acceleration fusion signals is proposed. Firstly, the surface electromyography signals of the rectus femoris, vastus medialis, tibialis anterior and gastrocnemius muscles and triaxial acceleration signals at the waist were collected as experimental data. Then the root mean square values of the surface electromyography and acceleration signals were extracted by sliding window method. Finally, a classifier based on support vector machine was constructed for the daily activities of the human body and the motion characteristics of the fall. Results: The experimental data totaled 320 sets of data, including 3 daily activities and forward fall, of which 160 sets of data were used as training sets and 160 sets of data were used as test sets. The recognition experiments were performed on four kinds of motions. The accuracy of the algorithm is 93.23%, the sensitivity is 92.4%, and the specificity is 100%, which achieves a good classification effect. Conclusion: The fall recognition algorithm based on SVM combined sEMG and acceleration has a good classification effect, which is of practical significance for the elderly fall protection. |
查看全文
查看/发表评论 下载PDF阅读器 |
关闭 |