Application Of Blind Source Separation In Vibration And Noise Analysis Of Machine Group

In implementing the condition monitoring and fault diagnosing of rotating machine groups, the measured both vibration and acoustic signals are influenced with each other, as a result they are instantaneous mixtures of themselves and noises. Therefore it is an essential step in the implementation to separate the individual signals of each machine from the signal mixtures without knowing the characteristics of the individual signals. This problem is the topic of blind source separation (BSS).In this paper the theory of BSS is first introduced and then it is applied to separate acoustic signals resulting from two different sources.In the application, the source separation is performed by means of both third order and fourth order cumulant methods. In this way the phase information of the signals is not lost. A simple and effective algorithm is developed, with which both third and fourth order cumulants will quickly approach zero. Numerical simulations and experimental verifications are carried out. The results show that the mixed signals can be significantly separated by BSS and this technique is promising to be used in condition monitoring and fault diagnosing of rotating machine groups.