Indexed by:Journal Papers

Date of Publication:2020-08-01

Journal:MECHANICAL SYSTEMS AND SIGNAL PROCESSING

Included Journals:EI、SCIE

Volume:142

ISSN No.:0888-3270

Key Words:FE model updating; Reduced-order models; Repeated eigenvalue structures; Incomplete measured modes

Abstract:In order to obtain a precise dynamic structural FE model for dynamic analysis, FE model updating is usually used to correct uncertainty parameters for an initial FE model using incomplete measured data. Despite numerous studies concerning FE model updating, the computational cost is still a challenging issue for the repeated eigenvalue structures. Firstly, an improved modal assurance criterion is proposed to evaluate the similarity of mode shapes for the repeated eigenvalue structures in this paper. And then, a novel ROM-based FE model updating framework consisting of an off-line phase and an on-line phase is proposed. In the off-line phase, a reduced-order basis is constructed by extracting primary components of a snapshot matrix using a proper orthogonal decomposition technique. The snapshot matrix represents a collection of static displacement vectors of the FE model under radial nodal loads, which are determined by incomplete measured mode shapes. In the on-line phase, FE model updating is performed via a reduced-order model with much cheaper computational cost. Finally, a numerical example and an experimental example demonstrate the accuracy and efficiency of the proposed framework. The results indicate that the proposed ROM-based FE model updating framework is more efficient and stable than the FOM-based FE model updating framework. (C) 2020 Elsevier Ltd. All rights reserved.