摘要
提出基于粒子群(PSO)优化支持向量机(SVM)的下肢假肢穿戴者跑动步态识别方法.将假肢接受腔装配的肌电(EMG)传感器、加速度计和足底的压力传感器采集的假肢穿戴者跑动运动信息进行去噪预处理,对应提取加速度的偏度、均值与肌电信号均方根多个特征参数作归一化处理,结合双下肢足底压力信息组成多维特征向量,作为SVM的输入,解决了单一特征识别步态的低准确率问题.利用PSO优化分类模型参数,建立基于SVM的次序二叉树分类模型对跑动步态进行辨识.与传统BP神经网络的步态识别方法对比表明,利用PSO优化SVM方法能够将跑动步态识别率提高到92.78%,优于SVM和BP神经网络.
An approach based on particle swarm optimization(PSO)and support vector machine(SVM)was proposed for running gait recognition of lower limb prosthesis wearers.The electromyography(EMG)sensors and accelerators installed in the prosthetic socket and pressure sensors installed in the plantar were used to acquire amputee’s running motion information,and the sensors data were denoised correspondingly.Then the skewness and mean of motion acceleration and the root-mean-square of EMG were chosen and normalized as feature parameters.These parameters were combined with the plantar pressure information to form multi-feature vector as the input of SVM,which solved the problem of low recognition accuracy of single feature.PSO was used to optimize classification model parameters.The binary tree model based on SVM was established to identify the running gait.The experimental results show that the recognition correct rate is 92.78%,which is higher than SVM and traditional BP neural networks.
作者
赵晓东
刘作军
陈玲玲
杨鹏
ZHAO Xiao-dong;LIU Zuo-jun;CHEN Ling-ling;YANG Peng(School of Control Science and Engineering,Hebei University of Technology,Tianjin 300130,China;Engineering Research Center of Intelligent Rehabilitation and Detecting Technology,Ministry of Education,Tianjin 300130,China)
出处
《浙江大学学报(工学版)》
EI
CAS
CSCD
北大核心
2018年第10期1980-1988,共9页
Journal of Zhejiang University:Engineering Science
基金
国家自然科学基金资助项目(61174009
61203323)
河北省青年自然科学基金资助项目(F2016202327)
河北省高等学校技术研究资助项目(ZC2016020)