Real time monitoring of muscular fatigue from dynamic surface myoelectric signals using a complex covariance approach.

A method aimed at the real-time monitoring of muscular fatigue was implemented and optimized. The method is based on an estimate of the complex covariance function in order to evaluate, in real time, the mean frequency of the myoelectric signal spectrum. Real-time implementation is guaranteed by a recursive computation of the complex covariance and then of the mean frequency. The results show good performance on both synthetic and experimental non-stationary myoelectric signals recorded during fatiguing dynamic protocols. Performance in the presence of noise is highly satisfactory on both deterministic signals and stochastic processes, even when there are strong non-stationarities. Moreover, the computational complexity is highly reduced with respect to that offered by traditional methods based on short time Fourier transform.