J Shanghai Jiaotong Univ Sci ›› 2021, Vol. 26 ›› Issue (1): 84-92.doi: 10.1007/s12204-020-2249-1

• • 上一篇    下一篇

Prosthetic Leg Locomotion-Mode Identification Based on High-Order Zero-Crossing Analysis Surface Electromyography

LIU Lei (刘磊), YANG Peng (杨鹏), LIU Zuojun (刘作军), SONG Yinmao (宋寅卯)    

  1. (1. College of Building Environment Engineering, Zhengzhou University of Light Industry, Zhengzhou 450002, China;
    2. College of Artificial Intelligence and Data Science, Hebei University of Technology, Tianjin 300130, China)
  • 出版日期:2021-02-28 发布日期:2021-01-19
  • 通讯作者: LIU Lei (刘磊) E-mail:liulei20060000@126.com

Prosthetic Leg Locomotion-Mode Identification Based on High-Order Zero-Crossing Analysis Surface Electromyography

LIU Lei (刘磊), YANG Peng (杨鹏), LIU Zuojun (刘作军), SONG Yinmao (宋寅卯)    

  1. (1. College of Building Environment Engineering, Zhengzhou University of Light Industry, Zhengzhou 450002, China;
    2. College of Artificial Intelligence and Data Science, Hebei University of Technology, Tianjin 300130, China)
  • Online:2021-02-28 Published:2021-01-19
  • Contact: LIU Lei (刘磊) E-mail:liulei20060000@126.com

摘要: The research purpose was to improve the accuracy in identifying the prosthetic leg locomotion mode. Surface electromyography (sEMG) combined with high-order zero-crossing was used to identify the prosthetic leg locomotion modes. sEMG signals recorded from residual thigh muscles were chosen as inputs to pattern classifier for locomotion-mode identification. High-order zero-crossing were computed as the sEMG features regarding locomotion modes. Relevance vector machine (RVM) classifier was investigated. Bat algorithm (BA) was used to compute the RVM classifier kernel function parameters. The classification performance of the particle swarm optimization-relevance vector machine (PSO-RVM) and RVM classifiers was compared. The BA-RVM produced lower classification error in sEMG pattern recognition for the transtibial amputees over a variety of locomotion modes: upslope, downgrade, level-ground walking and stair ascent/descent.


关键词: intelligent prosthesis, surface electromyography (sEMG), relevance vector machine(RVM), high-order
zero-crossing,
bat algorithm (BA), locomotion-mode identification

Abstract: The research purpose was to improve the accuracy in identifying the prosthetic leg locomotion mode. Surface electromyography (sEMG) combined with high-order zero-crossing was used to identify the prosthetic leg locomotion modes. sEMG signals recorded from residual thigh muscles were chosen as inputs to pattern classifier for locomotion-mode identification. High-order zero-crossing were computed as the sEMG features regarding locomotion modes. Relevance vector machine (RVM) classifier was investigated. Bat algorithm (BA) was used to compute the RVM classifier kernel function parameters. The classification performance of the particle swarm optimization-relevance vector machine (PSO-RVM) and RVM classifiers was compared. The BA-RVM produced lower classification error in sEMG pattern recognition for the transtibial amputees over a variety of locomotion modes: upslope, downgrade, level-ground walking and stair ascent/descent.


Key words: intelligent prosthesis, surface electromyography (sEMG), relevance vector machine(RVM), high-order
zero-crossing,
bat algorithm (BA), locomotion-mode identification

中图分类号: