采用小波变换和高斯过程的肌电信号模型预测

根据表面肌电信号的生物电信号特点,采用小波变换和高斯过程建模的方法对表面肌电信号进行建模和预测.对非线性的表面肌电信号利用拟合能力强大的高斯过程进行建模,预测效果较好,但所需运算时间长.针对其运算时间长的缺点进行改进,将预处理后的表面肌电信号小波分解,对分解后的系数高斯建模,然后重构.实验结果表明:该改进方法在响应时间和预测误差方面效果明显.

Model Forecasting of EMG Using Wavelet Transformation and Gaussian Process

According to the characteristics of the surface EMG signal, this paper uses wavelet transform and Gauss process modeling method to model and predict the surface EMG signal. The nonlinear surface EMG signal is used to model the fitting ability of the Gauss process, and the prediction effect is better, but the operation time is longer. To overcome the shortcomings of the long computation time, the wavelet decomposition of the surface EMG signal is processed, and the coefficients of the decomposition are modeled in Guassian. Experimental results show that the improved method has obvious effect on response time and prediction error.