外骨骼系统中控制信号的分析与处理
摘要
本文的主要工作是表面肌电信号采集和处理,及其在外骨骼系统中作为控制信号的应用,主要用于控制手臂和腿部的运动。表面肌电信号是由表面电极检测和记录的由神经肌肉活动产生的生物电信号。表面肌电信号反映肌肉的功能状态。因此,选择表面肌电信号作为控制信号具有直接、自然的特点。实验中的表面肌电信号是由加拿大Thought Technology Ltd.公司推出的MYOTRAC INFINITI Clinical T9850US肌电信号采集设备获得。
由于表面肌电信号信息量大,对采集的时间序列信号直接进行分类是不切实际的。因此需要提取信号的特征矢量。实验采集的表面肌电信号主要运用小波变换方法进行处理。然后提取肌电信号的有效特征值,结合BP神经网络进行分类,可以实现腿部不同运动模式的有效识别。此外,我们分析研究了步态周期内表面肌电信号和脚底压力信号的变化规律。基于表面肌电信号的特征矢量,选择支持向量机分类器实现了不同路况的识别。同时,利用ADAMS软件仿真人体步行,获得步态周期内脚底压力信号及对应的左膝的角度信号。然后,利用仿真数据训练自适应神经模糊推理系统(ANFIS)得到了脚底压力信号与人体步行时关节角度间的对应关系。
This thesis is concerned with the acquisition and processing of surface electromyogram (sEMG) signals for use in the controls of exoskeleton arm and leg motions. The sEMG signals are recorded by electrodes affixed to the skin surface and they capture the biological signals of the activities of the neuromuscular system. The sEMG signals reflect the muscles functional state. Therefore, using sEMG signals to control the motion of exoskeletons is not only a direct but also, a natural approach. The sEMG signals are obtained using the MYOTRAC INFINTI Clinical T9850US EMG Acquisition Instrument from the Canadian Thought Technology Ltd.
Due to the large amount of sEMG signals, it is not practical to feed the time sequence directly into a classifier. Instead, they are mapped into a smaller dimension feature vector. The experimentally acquired sEMG signals are then processed via a wavelet transform method. To extract the effective features from the sEMG signals a pattern recognition method, together wih the BP neural network classifier are applied. In our work, we were able to realize the recognition of different leg movement patterns. Further, we analyzed the change in the regularity of the sEMG signals and plantar pressure signals (PPS) for one gait cycle. Based on the sEMG characteristic vector, we used the support vector machine (SVM) classifier to recognize the road profile. Then, we employed the ADAMS software to perform the dynamics simulation, and obtained the simulated PPS and relative joint angle associated with the human gait. Next, we used the simulation data to train the adaptive neuro-fuzzy inference system (ANFIS), and obtained the relationship between PPS and joint angle for human gait.
由于表面肌电信号信息量大,对采集的时间序列信号直接进行分类是不切实际的。因此需要提取信号的特征矢量。实验采集的表面肌电信号主要运用小波变换方法进行处理。然后提取肌电信号的有效特征值,结合BP神经网络进行分类,可以实现腿部不同运动模式的有效识别。此外,我们分析研究了步态周期内表面肌电信号和脚底压力信号的变化规律。基于表面肌电信号的特征矢量,选择支持向量机分类器实现了不同路况的识别。同时,利用ADAMS软件仿真人体步行,获得步态周期内脚底压力信号及对应的左膝的角度信号。然后,利用仿真数据训练自适应神经模糊推理系统(ANFIS)得到了脚底压力信号与人体步行时关节角度间的对应关系。
This thesis is concerned with the acquisition and processing of surface electromyogram (sEMG) signals for use in the controls of exoskeleton arm and leg motions. The sEMG signals are recorded by electrodes affixed to the skin surface and they capture the biological signals of the activities of the neuromuscular system. The sEMG signals reflect the muscles functional state. Therefore, using sEMG signals to control the motion of exoskeletons is not only a direct but also, a natural approach. The sEMG signals are obtained using the MYOTRAC INFINTI Clinical T9850US EMG Acquisition Instrument from the Canadian Thought Technology Ltd.
Due to the large amount of sEMG signals, it is not practical to feed the time sequence directly into a classifier. Instead, they are mapped into a smaller dimension feature vector. The experimentally acquired sEMG signals are then processed via a wavelet transform method. To extract the effective features from the sEMG signals a pattern recognition method, together wih the BP neural network classifier are applied. In our work, we were able to realize the recognition of different leg movement patterns. Further, we analyzed the change in the regularity of the sEMG signals and plantar pressure signals (PPS) for one gait cycle. Based on the sEMG characteristic vector, we used the support vector machine (SVM) classifier to recognize the road profile. Then, we employed the ADAMS software to perform the dynamics simulation, and obtained the simulated PPS and relative joint angle associated with the human gait. Next, we used the simulation data to train the adaptive neuro-fuzzy inference system (ANFIS), and obtained the relationship between PPS and joint angle for human gait.
引文
[1]H.Kazerooni, Jean-Louis Racine, Lihua Huang, and Rayan Steger. On the control of the Berkeley Lower Extremity Exoskeleton (BLEEX).Proceedings of the 2005 IEEE International Conference on Robotics and Automation, Barcelona, Spain, April 2005:4353～4360.
[2]Hiroaki Kawamoto, Suwoong Lee, Shigehiro Kanbe and Yoshihuki Sankai.Power assist method for HAL-3 using EMG-based feedback controller. In Proceedings of the IEEE International Conference on Systems, Man and Cybernetics.2003(2):1648～1653.
[3]http://www.8264.com/38846.html
[4]http://www.lockheedmartin.com/products/hulc/HULCPhoto3.html
[5]http://baike.baidu.com/view/2716485.html
[6]http://www.medipp.com
[7]崔建国.上肢表面肌电信号的处理与运动模式辨别方法研究：[博士学位论文]东北大学信息科学与工程学院,2006
[8]何庆华,吴宝明,彭承琳.表面肌电信号的分析与应用.国外医学生物医学工程分册,Vo1.23,No.5,2000：299-303.
[9]Constable R, Thornhill RJ, Carpenter DR. Time-frequency analysis of SEMG during maximum height jumps under altered conditions [J]. Biomedical Sciences Instrumentation,1994(30):69～74.
[10]Maes F, Collign A. Multimodality image registration by maximization of mutual information [J]. IEEE Trans Med Imag,1997,16(2):187-198.
[11]姚喜妍.小波分析中的框架理论发展综述.运城学院学报.2005年10月,23(5)：7～9
[12]C. Valens. A really friendly guide to wavelets.1999
[13]王刚.基于小波变换和多重分形分析的表面肌电信号分析：[博士论文]上海交通大学.2008.
[14]陈胤,刘加海.表面肌电信号去噪研究.计算机时代[J].2008(6)：22～24
[15]赵敏.FFT变换与小波变换在变压器局部放电信号去噪中的应用.变压器.2009年5月.46(5)：28～31
[16]http://mpe.sjtu.edu.cn/UsersFiles/33/File/chapter-3_slides_26395.pdf
[17]陈香,杨基海,梁政,郑凡,钱晓进,冯焕清.NEMG信号运动单位动作电位的分类研究.中国科学技术大学学报.2003年8月.33(4)：466～472
[18]李青青.疲劳和步行的表面肌电图研究：[硕士论文]第三军医大学.2005年
[19]张立勋,王令军,王凤良,王克宽,高峻.步态训练机器人人机系统动力学仿真.高技术通讯.2009年第19卷第11期：1153～1158
[20]芮璋现,肖海波.支持向量机(SVM)及其应用.福建电脑.2007年第4期
[21]窦智宙,孟昭进.基于MTALAB的一对一M-SVM算法实现.山西煤炭管理干部学院学报.2008年1月
[22]吴晓莉,林哲辉等编著MATLAB辅助模糊系统设计.西安电子科技大学出版社.2002年8月
[23]张浩炯,余岳峰,王强.应用自适应神经模糊推理系统(ANFIS)进行建模与仿真.计算机仿真.2002年7月.19(4)：47～49
[24]顾秀萍.自适应神经模糊推理系统(ANFIS)及其仿真.火力与指挥控制.2010年2月.35 (2)
[25]王敬章,王人成,蔡付文,李芳,姜明文.表面机电信号的时域处理方法.2005国际康复论坛论文集—康复工程部分
[2]Hiroaki Kawamoto, Suwoong Lee, Shigehiro Kanbe and Yoshihuki Sankai.Power assist method for HAL-3 using EMG-based feedback controller. In Proceedings of the IEEE International Conference on Systems, Man and Cybernetics.2003(2):1648～1653.
[3]http://www.8264.com/38846.html
[4]http://www.lockheedmartin.com/products/hulc/HULCPhoto3.html
[5]http://baike.baidu.com/view/2716485.html
[6]http://www.medipp.com
[7]崔建国.上肢表面肌电信号的处理与运动模式辨别方法研究：[博士学位论文]东北大学信息科学与工程学院,2006
[8]何庆华,吴宝明,彭承琳.表面肌电信号的分析与应用.国外医学生物医学工程分册,Vo1.23,No.5,2000：299-303.
[9]Constable R, Thornhill RJ, Carpenter DR. Time-frequency analysis of SEMG during maximum height jumps under altered conditions [J]. Biomedical Sciences Instrumentation,1994(30):69～74.
[10]Maes F, Collign A. Multimodality image registration by maximization of mutual information [J]. IEEE Trans Med Imag,1997,16(2):187-198.
[11]姚喜妍.小波分析中的框架理论发展综述.运城学院学报.2005年10月,23(5)：7～9
[12]C. Valens. A really friendly guide to wavelets.1999
[13]王刚.基于小波变换和多重分形分析的表面肌电信号分析：[博士论文]上海交通大学.2008.
[14]陈胤,刘加海.表面肌电信号去噪研究.计算机时代[J].2008(6)：22～24
[15]赵敏.FFT变换与小波变换在变压器局部放电信号去噪中的应用.变压器.2009年5月.46(5)：28～31
[16]http://mpe.sjtu.edu.cn/UsersFiles/33/File/chapter-3_slides_26395.pdf
[17]陈香,杨基海,梁政,郑凡,钱晓进,冯焕清.NEMG信号运动单位动作电位的分类研究.中国科学技术大学学报.2003年8月.33(4)：466～472
[18]李青青.疲劳和步行的表面肌电图研究：[硕士论文]第三军医大学.2005年
[19]张立勋,王令军,王凤良,王克宽,高峻.步态训练机器人人机系统动力学仿真.高技术通讯.2009年第19卷第11期：1153～1158
[20]芮璋现,肖海波.支持向量机(SVM)及其应用.福建电脑.2007年第4期
[21]窦智宙,孟昭进.基于MTALAB的一对一M-SVM算法实现.山西煤炭管理干部学院学报.2008年1月
[22]吴晓莉,林哲辉等编著MATLAB辅助模糊系统设计.西安电子科技大学出版社.2002年8月
[23]张浩炯,余岳峰,王强.应用自适应神经模糊推理系统(ANFIS)进行建模与仿真.计算机仿真.2002年7月.19(4)：47～49
[24]顾秀萍.自适应神经模糊推理系统(ANFIS)及其仿真.火力与指挥控制.2010年2月.35 (2)
[25]王敬章,王人成,蔡付文,李芳,姜明文.表面机电信号的时域处理方法.2005国际康复论坛论文集—康复工程部分