Parallel Substructure Identification of Linear and Nonlinear Structures Using Only Partial Output Measurements

Abstract

Substructure identification with the idea of divide and conquer has played a significant role in the identification of large-scale structures. Nevertheless, most existing substructure identification methods are only used for linear structures. In addition, it is often required that the information of substructural interface forces are available or the interface responses are measured, which limits the application of substructure identification approaches in practice. In this paper, an improved substructure identification algorithm is proposed to identify linear/nonlinear structures and unknown inputs using only partial measurements of structural responses. In the proposed algorithm, the identification of substructural states and parameters including the parameters of the nonlinear models, unknown external excitations, and the substructural interface forces can be achieved without the measurement of substructural interface responses based on the generalized extended Kalman filter with unknown inputs, which was recently proposed by the authors. The proposed algorithm can identify each substructure in parallel because no information needs to be transferred between the adjacent substructures. Some numerical identifications of linear and nonlinear structures were conducted to demonstrate the proposed substructure identification algorithm and verify the efficiency for the identification of the large-scale structures.