基于GMM-HMM模型的智能下肢假肢运动意图识别  

Motion intent recognition of intelligent lower limb prosthesis based on GMM-HMM

作　　者：盛敏 刘双庆 王婕 苏本跃 Sheng Min;Liu Shuangqing;Wang Jie;Su Benyue(School of Mathematics and Computational Science, Anqing Normal University, Anqing 246133, China;University Key Laboratory of Intelligent Perception and Computing of Anhui Province, Anqing 246133, China;School of Computer and Information, Anqing Normal University, Anqing 246133, China)

机构地区：[1]安庆师范大学数学与计算科学学院,安庆246133 [2]安徽省智能感知与计算重点实验室,安庆246133 [3]安庆师范大学计算机与信息学院,安庆246133

出　　处：《仪器仪表学报》2019年第5期169-178,共10页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(11471093);教育部“云数融合科教创新”基金(2017A09116);安徽省科技重大专项(18030901021);安徽省高校优秀拔尖人才培育项目(gxbjZD26)资助

摘　　要：传统下肢假肢运动意图识别方法常使用多模态传感器信号,带来一定的复杂性以及模式转换识别一般带有滞后性,提出了基于数据驱动下的智能下肢假肢运动意图识别方法。在对单侧下肢截肢者运动模式进行了重定义后,仅使用惯性传感器,采集健肢侧处于摆动相的时序数据。选择高斯混合-隐马尔可夫模型作为分类器,对下肢假肢的运动意图进行识别。实验结果表明,该算法在模式空间中的一组基模式:平地行走、上坡、下坡、上楼和下楼5种稳态模式中,识别率达到98.99%,在包含5种稳态模式和8类转换模式的13类运动模式中的识别率可达到96.92%。所提出的方法可以在下肢假肢运动意图识别性能上有较大提升,帮助单侧下肢截肢者实现自然、流畅、稳定的行走。Traditional lower limb prosthesis motion intent recognition methods often use multi-modal sensor signals, which bring certain complexity and lags of pattern transition recognition. This paper proposes a data-driven based intelligent lower limb prosthesis motion intent recognition method. After redefining the movement patterns of unilateral lower limb amputees, only the inertial sensorsare used to collect the time series data in the swing phases of the healthy side. The Gaussian mixture model-hidden Markov model (GMM-HMM) is selected as the classifier to recognize the motion intent of lower limb prosthesis. The experiment results show that the recognition rate of the method reaches 98. 99% in steady patterns: level-ground walking, ramp ascent, ramp descent, stair ascent and stair descent, and 96. 92% in 13 motion patterns that contain 5 steady patterns and 8 transition patterns. The method proposed in this paper can greatly improve the recognition performance of lower limb prosthesis motion intent, and help the unilateral lower limb amputees to walk naturally, smoothly and steadily.

关 键 词：运动意图识别 惯性传感器 高斯混合-隐马尔可夫模型 模式转换 摆动相 

分 类 号：TH7[机械工程—仪器科学与技术] TP391[机械工程—精密仪器及机械]