基于多源生物信号的下肢步态相识别  

Gait Phase Recognition Based on Multi-source Biological Signals

作　　者：张启忠[1] 席旭刚[1] 罗志增[1] ZHANG Qi-zhong;XI Xu-gang;LUO Zhi-zeng(Intelligent Control ＆ Robotics Institute,Hangzhou Dianzi University,Hangzhou,Zhejiang 310018,China)

机构地区：[1]杭州电子科技大学智能控制与机器人研究所,浙江杭州310018

出　　处：《计量学报》2018年第6期895-901,共7页Acta Metrologica Sinica

基　　金：国家自然科学基金(61671197);浙江省自然科学基金(LY17F030021);浙江省公益技术研究项目(LGF18F010006)

摘　　要：为提高人体下肢步态相识别的准确性,研究了融合表面肌电信号(s EMG)、膝关节角度和足底压力信号的人体下肢步态相识别方法。首先,将s EMG信号进行小波包分解提取多尺度能量和多尺度模糊熵特征;然后,对提取的s EMG信号特征值采用主成分分析(PCA)方法进行降维处理,并与足底压力特征值和膝关节能量特征值构成一组特征向量;最后,将特征向量输入粒子群优化最小二乘支持向量机(PSO-LSSVM)对人体下肢运动信息进行步态相识别。实验结果表明,所提方法相较于其他方法有较高的识别准确率和有效性。In order to improve the accuracy of gait recognition, the gait recognition of human lower limb was studied based on the fusion of surface electromyography （ sEMG）, knee joint angle and plantar pressure. Firstly, the sEMG signals were decomposed by wavelet packet to extract the features of multi-scale energy and multi scale fuzzy entropy. Then, the principal component analysis （PCA） method was employed to reduce the dimension of the feature value of sEMG, and the feature vectors were constituted by the features of sEMG, plantar pressure and the knee energy. Finally, the feature vectors were inputted into the least squares support vector machine （PSO-LSSVM） optimized by the particle swarm to recognize gait of lower limb. The experimental results show that this method has higher recognition accuracy and validity than other methods.

关 键 词：计量学 肌电信号 步态相识别 特征提取 粒子群优化 最小二乘支持向量机 模式识别 

分 类 号：TB973[一般工业技术—计量学]