Quantitative relationship modeling between Surface Electromyography and elbow joint angle

Research based on Surface Electromyography (sEMG) has important significance in enlightening the control precision of artificial limb. sEMG and elbow joint angle from 4 muscles of human´s upper limb is extracted in the process of elbow bending and stretching. The quantitative relationship model of sEMG and elbow joint angle is established through Nerve Network (sEMG feature as input set and elbow joint angle as output set) and the prediction effect is also estimated based on two methods (Mean value and RMS of the error between predicted joint angle and actual joint angle; linear regressive method). Prediction comparison between different feature extraction methods (RMS and Mean Power Frequency) is also analyzed. The experimental results show that the mean error between the prediction angle based on sEMG and the practical angle are all less than 1 degree which means good prediction effect.