Blind sources separation of harmonic signals for output only modal analysis

• Autor(a) principal: Li, Xinhui
• Data de Publicação: 2013
• Outros Autores: Brennan, Michael J. [UNESP], Yang, Tiejun, Lu, Zeqi, Dong, Jianchao
• Tipo de documento: Artigo de conferência
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://hdl.handle.net/11449/227803
• topic: BSS; FastICA; SOBI

Blind source separation (BSS) is an emerging technique in array processing and data analysis, aimed at recovering unobserved sources from observed signals which are a filtered mix of the sources. The only assumption in the technique is the mutual independence or uncorrelation of the source signals. This technique has gained recent interest in modal analysis. As one of the widely used techniques for blind source separation, independent component analysis (ICA) fails when the frequency ratio of two harmonic sources is close to three. This is a disadvantage of ICA when it is used to do modal analysis. In the present study, an analytical method is employed to find out why the technique fails. An alternative technique, namely secondary-order blind identification (SOBI) is shown to avoid such a problem. Both of the two techniques are compared by way of numerical simulations and experiments. Copyright© (2013) by Austrian Noise Abatement Association (OAL).