Abstract:

A method integrated by independent component analysis（ICA） and empirical mode decomposition （EMD） is proposed, which can eliminate the white noise from different signals. The superiority of ICA and EMD is expatiated emphatically. The basic theory and the detailed courses of construction of virtual noise channel used EMD are described. Based on time-frequency spectral characteristic of white noise, the principle of select intrinsic mode function(IMF), used the Hilbert time-frequency spectra，is proposed to reconstruct virtual noise channel. The simulations indicate that the denoising performance is effective. Compared with the denoising performance, this method is better than traditional wavelet method.

Key words: Empirical Mode Decomposition, Independent component analysis, White Noise, Virtual Noise Channel, Denoising