Separation of sinusoidal sources

A particular source separation problem is addressed in this paper. We mainly focus on the separation of convolutive mixtures of rotating machine noises when the rotation speeds are close. Three specific points are developed. In the first point, we study the feasibility of the separation of periodic signals, regarding the hypothesis of random and non gaussian sources. We also discuss about the hypothesis of independence between the sources, as a function of the rotation speeds. In the second point, we analyze the performances of the source separation for close rotation speeds. They are linked to a partial correlation between the machine noises. Then, we propose a new method for very close rotation speeds, which takes into account this partial dependence between the sources.