Estimation of muscle fatigue using wavelet decomposition

Muscle fatigue causes numerous injuries among workers in industries involving mechanical labour each year. This study explores a new feature based on energy of detail wavelet coefficient for muscle fatigue detection in the upper limb. The muscle fatigue data was generated after isometric muscle action. 7 subjects underwent 3 trails each using 3 channels corresponding to Biceps Brachii, Extensor Digitorum Communis and Flexor Carpi Radialis muscle respectively. It was found that the energy of detail coefficients of 3^rd, 4^th and 5^th level of wavelet decomposition (15.625-125 Hz) increases as the muscle fatigue level increases. Moreover, it was also found that the intercept of regression curve plotted to approximate this increase and the rate of increase of the energy, had maximum value for the energy of 3^rd level of wavelet decomposition followed by 4^th and 5^th level resp. To detect the onset of fatigue, the energy of coefficients of frequency 15.6-62.5 Hz at the start and end of duration of time for which the isometric contraction is performed is considered. The fatigue was detected among subjects when this average final energy value was approximately five times the initial value for 15.6-62.5 Hz range.