Modeling and classification of sEMG based on blind identification theory

Abstract

Surface electromyography signal is non-stationary, susceptible to external interference. For this situation under this case, cyclostationary input with the inverse nonlinear mapping of the Hammerstein-Wiener model were combined to build surface electromyography model and to realize the blind discrete nonlinear system identification. The parameters of model were used as input of improved BP neural network. The experiments results demonstrated the effectiveness of this approach. © 2011 Springer-Verlag.