Research and Application of Noise Blind Mixed Signal On-line SeparationAlgorithm Based on ICA

摘要

Independent Component Analysis (ICA) is a kind of novel multi-dimensional statistical analysis method. Itscharacteristic is to recover the independent source signal components from the multi-channel observation signal. In thispaper, based on the model of independent component analysis, we used the gradient style algorithm to separate the noiseblind mixed electrical signal. In the application of gradient algorithm, in order to solve the compromise between convergencerate and stability performance of the natural gradient blind source separation algorithm, we adopted the variablestep size natural gradient algorithm to complete the noise electrical signal on-line blind source separation simulation,which provides an effective method to realize the multi-sensor collection signal.