An Improved Blind Source Separation for Structural Mode Identification Using Fewer Measurements

Abstract

In this paper, we present a novel underdetermined structural mode identification algorithm within the framework of Blind Source Separation (BSS) employing PARAllel FACtor (PARAFAC) decomposition. In the proposed method, BSS is employed on vibration measurements to generate an over-complete dictionary of multi-component bases, which are then used to estimate the modal parameters by utilizing the strong uniqueness property of PARAFAC decomposition. Results show significant computational efficiency of the proposed method over existing BSS algorithms, while substantially improving the upper bound of source separation capability of existing PARAFAC decompositions.