sEMG时频特征线性回归法与非线性神经网络法预测伸膝肌群极限功率保持能力测试中功率损失率的比较研究

发布时间：2019-08-02 15:02

【摘要】：目的:拟比较s EMG时频特征线性回归法与非线性神经网络法预测伸膝肌群极限功率保持能力测试中功率损失率的差异。方法:BTE Primus~(RS)系统与肌电仪同步,40名男大学生膝关节重复性屈伸运动至疲劳,阻力设置50%等长峰值力矩,动作频率60次/min。求取每次伸膝阶段极限功率损失率(Power%),伸膝肌群sEMG时域(MAV%、RMS%)、频域(MNF%、MDF%)与瞬时频率(IMNF%、IMDF%)参数变化率,基于s EMG时频特征参数(MAV、ZC、SSC、WL)建立多层感知人工神经网络模型,求取功率真实值与估计值。结果:IMDF%能单独解释股内肌、股直肌与股外肌极限功率损失率的方差变异为6.33%、22.71%、12.31%,IMDF%联合其他时频参数一起能解释的方差变异为6.95%、25.93%和16.05%,非线性神经网络法求取的功率估计值能解释的方差变异为10.43%、34.23%和18.05%,且信噪比值逐步增大。线性与非线性技术功率真实值与估计值拟合所得两直线的斜率与截距有显著性差异(P0.05)。结论:s EMG时频特征线性回归法与非线性神经网络法,均能很好地追踪人体神经肌肉系统动态工作疲劳过程中输出功率的损失,但后者的准确性要优于前者。
[Abstract]:Aim: to compare the difference of power loss rate between s EMG time-frequency characteristic linear regression method and nonlinear neural network method in predicting the ultimate power retention ability of knee extensor muscle group. Methods: BTE Primus~ (RS) system synchronized with EMG. 40 male college students had repetitive flexion and extension movements to fatigue. The resistance was set at 50% equal length peak torque, and the action frequency was 60 times / min.. The limit power loss rate (Power%), sEMG time domain (MAV%,RMS%), frequency domain (MNF%,MDF%) and instantaneous frequency (IMNF%,IMDF%) parameters of each knee extension stage were obtained. A multi-layer perceptual artificial neural network model was established based on sEMG time-frequency characteristic parameters (MAV,ZC,SSC,WL), and the real and estimated values of power were obtained. Results: the variance variation of limit power loss rate of medial muscle, rectus femoris and extrathigh muscle was 6.33%, 22.71% and 12.31%, respectively. The variance variation explained by IMDF% combined with other time-frequency parameters was 6.95%, 25.93% and 16.05%. The variance variation of power estimation obtained by nonlinear neural network method was 10.43%, 34.23% and 18.05%, respectively. the variance variation of IMDF% combined with other time-frequency parameters was 6.95%, 25.93% and 16.05%, and the variance variation of IMDF% combined with other time-frequency parameters was 6.95%, 25.93% and 16.05%. The signal-to-noise ratio (SNR) increases gradually. The slope and intercept of the two lines fitted with the real and estimated power of linear and nonlinear techniques are significantly different (P 0.05). Conclusion both: s EMG time-frequency characteristic linear regression method and nonlinear neural network method can track the loss of output power in the dynamic fatigue process of human neuromuscular system, but the accuracy of the latter is better than that of the former.
【作者单位】：东北师范大学;长春光华学院商学院;北京航空航天大学生物医学工程学院;中国标准化研究院;
【基金】：科技基础性工作专项(2013FY110200) 中央高校基本科研业务费资助项目(14QNJJ032)
【分类号】：R318.01

【相似文献】