Blind Source Separation of Mechanical Vibration Signal in Time Domain
Li Shunming
(College of energy and power engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016)

Abstract: The basic principle of blind source separation is introduced, and the characteristics of the instantaneous mixing model and the convolutive mixing model are analyzed. Two independence criteria of source signal are proposed simulta neously. The estimate error function of correctness in blind source separation is researched. For real mechanical vibration signal, the convolutive instantaneous mixing model in time domain is used for blind separation of source signal, and a satisfied result is achieved. The astringency of the method in blind separat ion to multi-source signal is discussed finally.
Keywords: mechanical vibration, signal, blind source separation.