Adaptive estimation of spectral parameters for myoelectric signals by means of AR modelling

The authors report the results obtained using an autoregressive (AR) method in the study of myoelectric signals during sustained contractions, when fatigue is supposed to occur. The differences between this sequential method and a batch Burg method are reported. Good results were obtained when the method was applied to experimentally recorded signals, and the computational burden was acceptable. The adaptive algorithm used makes it possible to estimate spectral parameters almost continuously and to control the variance of the estimates