Blind Source Separation of Cyclo-stationary Signals Using an Extended Natural Gradient Method and Its Application
Ke Wei · Ship & Ocean Engineering · 2011
The method of blind source separation of signal about cyclo-stationary was derived by applying natural gradient iterative learning to the novel cost function defined according to instantaneous mixtures.The efficiency of the algorithm was supported by simulations,which showed that the proposed method has improved performance for the separation in terms of waveform similarity and convergence speed,as compared to the traditional ICA method.In the fault rolling element bearing experiment,fault signal is extracted effectively.