基于小波变换的肌肉疲劳表面肌电信号特征提取的研究
|
摘要
当前肌肉疲劳表面肌电信号(surface electromgography,sEMG)特征提取方法,忽略了非线性跳错信号的影响,且不能在非平稳状态下进行特征提取,存在特征提取准确度差的问题。提出基于小波变换的肌肉疲劳sEMG特征提取研究,采用小波变换对所采集的样本去噪,结合时域、频域特征分析法,融合傅里叶变换方法对肌电信号中的线性特征进行提取,根据带谱近似熵理论对非线性挑错信号进行特征回归分析,并利用拟态分解函数和希尔伯特变换法对肌电信号进行时频特征的整合提取,最终完成基于小波变换的肌肉疲劳sEMG特征提取研究。实验验证,所提方法具有可行性,且将1000个肌电信号样本分成5组,对其中的跳错信号进行特征提取,所提方法准确度较文献方法高出75%,在非平稳状态下将200个肌电信号样本分成5组进行特征提取,所提方法准确度较文献方法高出33%。由此得出,所提方法优于当前特征提取方法。
At present, the feature extraction method of muscle fatigue surface electromyogram(MSEMG) signal ignores the influence of non-linear error-jumping signal, and can not extract feature in non-stationary state, which causes poor accuracy of feature extraction.To propose the feature extraction of muscle fatigue surface electromyogram signal based on wavelet transform. Wavelet transform was used to denoise the collected samples. The linear feature of EMG signal was extracted by combining time-domain and frequency-domain feature analysis with Fourier transform method. According to band-spectrum approximate entropy theory, the feature regression analysis of non-linear fault-detection signal was carried out, and the pseudo-decomposition function and Hirsch function were used. The time-frequency features of EMG signals were integrated and extracted by the method of Albert transform. Finally, the feature extraction of muscle fatigue surface EMG signals based on wavelet transform was completed. Experiments showed that the proposed method was feasible, and 1 000 EMG samples were divided into five groups to extract the feature of the error signal. The accuracy of the proposed method was 75% higher than that of the literature method. In the non-stationary state, 200 EMG samples were divided into five groups to extract the feature. The accuracy of the proposed method was 33% higher than that of the literature method. It is concluded that the proposed method is superior to the current feature extraction method.
At present, the feature extraction method of muscle fatigue surface electromyogram(MSEMG) signal ignores the influence of non-linear error-jumping signal, and can not extract feature in non-stationary state, which causes poor accuracy of feature extraction.To propose the feature extraction of muscle fatigue surface electromyogram signal based on wavelet transform. Wavelet transform was used to denoise the collected samples. The linear feature of EMG signal was extracted by combining time-domain and frequency-domain feature analysis with Fourier transform method. According to band-spectrum approximate entropy theory, the feature regression analysis of non-linear fault-detection signal was carried out, and the pseudo-decomposition function and Hirsch function were used. The time-frequency features of EMG signals were integrated and extracted by the method of Albert transform. Finally, the feature extraction of muscle fatigue surface EMG signals based on wavelet transform was completed. Experiments showed that the proposed method was feasible, and 1 000 EMG samples were divided into five groups to extract the feature of the error signal. The accuracy of the proposed method was 75% higher than that of the literature method. In the non-stationary state, 200 EMG samples were divided into five groups to extract the feature. The accuracy of the proposed method was 33% higher than that of the literature method. It is concluded that the proposed method is superior to the current feature extraction method.
引文
[1]王蒙,姚海霞.基于表面肌电的运动员上肢肌疲劳特征分析[J].生物医学工程研究,2018,37(4):487-491.
[2]章亚男,景银平,沈林勇,等.下肢表面肌电信号的降维和映射分析[J].传感技术学报,2018,31(7):1046-1053.
[3]夏培玲,刘昕,姜广富.表面肌电信号提取方法及其在体育领域中的医药应用[J].当代体育科技,2018,8(19):6-7.
[4]李琰,隋修武.基于平移不变小波变换的颈肩肌电信号去噪方法研究[J].生物医学工程研究,2018,37(1):52-55.
[5]刘明君,董增寿.基于改进小波变换的手臂肌电信号去噪算法的研究[J].电子技术应用,2018,44(3):122-125.
[6]王光旭,崔建松,张庆光,等.采用优化神经网络的下肢肌电信号的识别[J].厦门理工学院学报,2018,26(1):55-59.
[7]高云园,任磊磊,周旭,等.基于变尺度符号传递熵的多通道脑肌电信号耦合提取[J].中国生物医学工程学报,2018,37(1):8-16.
[8]魏鹏娜,李创,唐荣年.用于脑瘫儿下肢康复装置的表面肌电信号特征提取方法研究[J].生物医学工程研究,2017,36(3):229-233.
[9]许全盛,李世明,季淑梅.基于表面肌电信号的膝关节动态力矩的BP神经网络估计[J].中国体育科技,2018,54(1):105-114.
[10]王坤朋,庞杰,石磊,等.人体腿部表面肌电信号特征提取方法[J].重庆大学学报,2017,40(11):83-90.
[11]高云园,任磊磊,张迎春,等.基于相干性的多频段脑肌电信号双向耦合分析[J].传感技术学报,2017,30(10):1465-1471.
[12]顾明亮,刘俊.改进的PSO-SVM在表面肌电信号模式识别中的研究[J].传感技术学报,2017,30(10):1459-1464.
[2]章亚男,景银平,沈林勇,等.下肢表面肌电信号的降维和映射分析[J].传感技术学报,2018,31(7):1046-1053.
[3]夏培玲,刘昕,姜广富.表面肌电信号提取方法及其在体育领域中的医药应用[J].当代体育科技,2018,8(19):6-7.
[4]李琰,隋修武.基于平移不变小波变换的颈肩肌电信号去噪方法研究[J].生物医学工程研究,2018,37(1):52-55.
[5]刘明君,董增寿.基于改进小波变换的手臂肌电信号去噪算法的研究[J].电子技术应用,2018,44(3):122-125.
[6]王光旭,崔建松,张庆光,等.采用优化神经网络的下肢肌电信号的识别[J].厦门理工学院学报,2018,26(1):55-59.
[7]高云园,任磊磊,周旭,等.基于变尺度符号传递熵的多通道脑肌电信号耦合提取[J].中国生物医学工程学报,2018,37(1):8-16.
[8]魏鹏娜,李创,唐荣年.用于脑瘫儿下肢康复装置的表面肌电信号特征提取方法研究[J].生物医学工程研究,2017,36(3):229-233.
[9]许全盛,李世明,季淑梅.基于表面肌电信号的膝关节动态力矩的BP神经网络估计[J].中国体育科技,2018,54(1):105-114.
[10]王坤朋,庞杰,石磊,等.人体腿部表面肌电信号特征提取方法[J].重庆大学学报,2017,40(11):83-90.
[11]高云园,任磊磊,张迎春,等.基于相干性的多频段脑肌电信号双向耦合分析[J].传感技术学报,2017,30(10):1465-1471.
[12]顾明亮,刘俊.改进的PSO-SVM在表面肌电信号模式识别中的研究[J].传感技术学报,2017,30(10):1459-1464.