A super-element approach for structural identification in time domain

Abstract

For most time-domain identification methods, a complete measurement for unique identification results is required for structural responses. However, the number of transducers is commonly far less than the number of structural degrees of freedom (DOFs) in practical applications, and thus make the time-domain identification methods rarely feasible for practical systems. A super-element approach is proposed in this study to identify the structural parameters of a large-scale structure in the time domain. The most interesting feature of the proposed super-element approach is its divide-and-conquer ability, which can be applied to identify large-scale structures using a relatively small number of transducers. The super-element model used for time domain identification is first discussed in this study. Then a parameterization procedure based on the sensitivities of response forces is introduced to establish the identification equations of super-elements. Some principles are suggested on effective decomposing of the whole structure into super-elements for identification purposes. Numerical simulations are conducted at the end of this study. The numerical results show that all structural parameters can be identified using a relatively small number of transducers, and the computational time can also be greatly shortened.