BLIND SEPARATION OF NOISY HARMONIC SIGNALS USING ORTHOGONAL TECHNIQUES FOR ROTATING MACHINE DIAGNOSIS
P. Fabry, Ch. Servi`ere
email : Christine.Serviere@lis.inpg.fr
We present a robusttonoise technique for the blind
source separation of harmonic signals, using temporal
correlations. The mixtures are convolutive and cor
rupted with spatially correlated noises. Their distribu
tions and correlation functions are unknown. As the
usual Principal Components Analysis cannot provide
a correct estimate of the signal subspace in this situa
tion, we introduce a new estimator of the decorrelation
matrix and the signal subspace. It exploits the model
of harmonic sources and uses nonhermitian interspec
tral matrices of delayed observations. After whitening,
the delayed observations are used again to determine
the Givens plane rotations that complete the separa
tion process. A Wiener filtering is then applied for
denoising the outputs. The proposed method is effi
cient for low Signal to Noise Ratio and illustrated with
simulations and rotating machine vibration data.