文章摘要
刘磊,杨鹏,刘作军,宋寅卯.基于多核学习极限学习机的助行机器人运动相容性识别[J].高技术通讯(中文),2021,31(9):978~985
基于多核学习极限学习机的助行机器人运动相容性识别
Motion compatibility recognition of walking robot based on multi-core learning limit learning machine
  
DOI:10.3772/j.issn.1002-0470.2021.09.009
中文关键词: 多核学习; 极限学习机(ELM); 表面肌电信号(sEMG); 运动相容性识别
英文关键词: multi-core learning, extreme learning machine (ELM), surface electromyography (sEMG), motion-compatibility recognition
基金项目:
作者单位
刘磊  
杨鹏  
刘作军  
宋寅卯  
摘要点击次数: 814
全文下载次数: 526
中文摘要:
      下肢外骨骼作为一种可穿戴设备可以保护人体、增强人体的力量、激发人体的自我修复能力,已经在康复领域广泛使用。由于外骨骼机器人实际运动意图与期望运动意图存在一定差异(即运动不相容),容易造成穿戴者不舒适。为了提高运动相容性识别的准确率,本文针对步幅过大、步幅过小、步幅相容采用多核学习极限学习机(ELM)的方法进行了识别。表面肌电信号(sEMG)包含了大量的步态信息,能够很好地应用于运动相容性的识别,首先从时域、频域、时频域等角度提取了表面肌电信号特征,然后利用灰狼算法优化极限学习机核函数参数,最后用多核极限学习机理论,获得最优的分类模型。实验结果表明,基于多核学习极限学习机的助行机器人运动相容性识别准确率较单核极限学习机有明显提高。
英文摘要:
      Lower limb exoskeleton as a wearable device can protect the human body, strengthen the power of human body, and stimulate the body’s ability to repair itself. It has been widely used in the field of rehabilitation. Because the robot motion intention and actual expected intention of the exoskeleton robot exist certain differences (incompatible), it is easy to cause uncomfortable of the wearers. In order to improve the recognition accuracy of motion compatibility, a method of multi-core learning extreme learning machine (ELM) is used for large stride length, small stride length, and compatible stride length. Surface electromyography (sEMG) signal contains a large amount of gait information, which is applied to the recognition of motion compatibility. Firstly, the features of sEMG signals are extracted from time domain, frequency domain, and time frequency domain. Then, the extreme learning machine kernel function parameters are optimized by the gray wolf algorithm with multi-core extreme learning machine theory. Finally, the optimal classification model is obtained by using multi-core learning extreme learning machine theory. The experimental results show that the recognition accuracy based on multi-core learning extreme learning machine of walking robot motion compatibility recognition is better than extreme learning machine.
查看全文   查看/发表评论  下载PDF阅读器
关闭

分享按钮