EMG based muscle fatigue detection using autocorrelation and k-means clustering

Abstract

The electromyogram (EMG) can be commonly used to detect muscle fatigue during exercise to prevent injury or muscle disorder. A traditional way to indicate fatigue is based on extracting features from EMG segments in time-domain or frequency-domain. In this work, muscle fatigue detection is developed by extracting three features from the autocorrelation function of EMG segments. The classification is done using a k-means clustering approach. The proposed method has also successfully classified unknown EMG segments into non-fatigue and fatigue state. The accuracy of the proposed method is evaluated in detecting the signal of transition-to-fatigue stage.