Characterization of Muscle Fatigue Degree in Cyclical Movements Based on the High-Frequency Components of sEMG.

Prolonged and high-intensity human-robot interaction can cause muscle fatigue. This fatigue leads to changes in both the time domain and frequency domain of the surface electromyography (sEMG) signals, which are closely related to human body movements. Consequently, these changes affect the accuracy and stability of using sEMG signals to recognize human body movements. Although numerous studies have confirmed that the median frequency of sEMG signals decreases as the degree of muscle fatigue increases-and this has been used for classifying fatigue and non-fatigue states- there is still a lack of quantitative characterization of the degree of muscle fatigue. Therefore, this paper proposes a method for quantitatively characterizing the degree of muscle fatigue during periodic exercise, based on the high-frequency components obtained through ensemble empirical mode decomposition (EEMD). Firstly, the sEMG signals of the estimated individuals are subjected to EEMD to obtain the high-frequency components, and the short-time Fourier transform is used to calculate the median frequency (MF) of these high-frequency components. Secondly, the obtained median frequencies are linearly fitted, and based on this, a standardized median frequency distribution range (SMFDR) of sEMG signals under muscle fatigue is established. Finally, a muscle fatigue estimator is proposed to achieve the quantification of the degree of muscle fatigue based on the SMFDR. Experimental validation across five subjects demonstrated that this method effectively quantifies cyclical muscle fatigue, with results revealing the methodology exhibits superiority in identifying multiple fatigue states during cyclical movements under consistent loading conditions.