Improvement of control of the compression of the spectral density of a surface EMG signal: application to fatigue

The nature of quasi-stationary surface EMG signals permits estimation of their spectral density by the periodogram method. During a sustained muscle contraction, the spectral density undergoes an compression towards the lower frequencies. The spectral displacement may be quantified by the average and median frequencies. According to the processes of fatigue, these characteristics obey either the law of linear regression or the law of exponential regression. The inconvenience of the two models of regression and the means of their avoidance are presented. We show in this article the limit of the previous solution independent of the type of regression and we propose the procedure which allows them to be overcome