A blind source separation based approach for modal parameter estimation in traditional input-output eperimental modal analysis framework (bn1724)

Though Blind Source Separation (BSS) techniques have typically been applied within Operational Modal Analysis (OMA) framework, there joint diagonalization property can also be utilized within traditional input-output Experimental Modal Analysis (EMA) for the purpose of modal parameter estimation. This paper explores this possibility and suggests an approach to achieve the same. The paper presents the mathematical concepts that make it possible for Second Order Blind Identification (SOBI), a blind source separation algorithm to be applied to impulse response functions for the purpose of modal parameter estimation. It is also shown in the paper that, in its current form, the approach provides real mode shapes, irrespective of the nature of the system under investigation. In addition to demonstrating the approach by means of studies conducted on an analytical 5 degrees-of-freedom system, the real mode shapes extracted using the suggested approach are also used for model updating purposes.