Segmenting method for surface electromyogram signal activity section based on sample entropy and Gaussian model

Abstract

The invention discloses a segmenting method for a surface electromyogram signal activity section based on sample entropy and a Gaussian model. The segmenting method comprises the steps that 1 electricpotential values of surface electromyogram signals are collected from the related muscles of the human body, widths u and step lengths f of sliding windows are set, and sliding segmentation is conducted on data; 2 sample entropy parameters m and r are set, and the sample entropy of each sliding window is calculated; 3 a DBSCAN cluster method is adopted for determining a Gaussian polynomial basicmodel, and then model parameters are initialized; 4 a nonlinear least square method is adopted for fitting of the Gaussian model; 5 according to the Gaussian model, an energy threshold is determined,and the activity section is segmented. According to the method, the characteristic can be utilized that the sample entropy has an inhibition effect on noise, the influence of the noise on signal segmentation is overcome, and therefore the foundation is provided for precision and reliability of further analysis for the surface electromyogram signals.

Description

technical field [0001] The invention belongs to the field of surface electromyography signal processing, in particular to a method for segmenting active segments of surface electromyography signals. Background technique [0002] Currently, there are three main types of active segment detection and segmentation methods for surface EMG signals: the first type is based on the method of setting thresholds based on signal amplitude, which is sensitive to noise and is not suitable for environments with low signal-to-noise ratios; To solve this problem, a method based on wavelet transform has been proposed, which can work when the wavelet model is compatible with the measured myoelectric motion signal, so it is necessary to find a suitable wavelet function, but in most cases it is difficult Find the ideal wavelet function. The second category is the analysis using the maximum likelihood ratio method in statistics, which has a complete theoretical derivation, but it needs to be est...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a method based on sample entropy and Gaussian in order to overcome the difficulties that the existing method cannot overcome the low signal-to-noise ratio, many restrictive conditions, and limited application range and cannot segment the active segment due to the detection of the active segment of the surface electromyography signal. The method of segmenting the active segment of the surface electromyographic signal of the model is expected to be able to segment the active segment of the surface electromyographic signal under various conditions, and overcome the influence of noise on the analysis of the surface electromyographic signal, so as to improve the accuracy and accuracy of the surface electromyographic signal analysis. Reliability provides the basis

Images