Auto-regression analysis upon EMG power spectra during dynamic exercise

Surface electromyograms were recorded as interferance signals, which contained many sorts of motor units during dynamic exercise. Those complex signals have been transformed to a simple power spectra, that have been used as the index of muscle fatigue or recruitment of the motor unit. This study applies the Auto-Regression Model (ARM) for those EMGs Power Spectra, and proposes a new classification of motor units. As the dynamic exercise, the cranking movement of the upper limb or the pedalling movement of the lower limb were used. At the result of analysis, EMGs power spectra were divided into three elements of the low (5-35 Hz), middle(36-70 Hz), and high(over 76 Hz) frequency bands with computer caliculation of ARM.