Output-Only Identification of Base Excited Structures Using Subspace Methods With Asymmetric Partitions

Output-only identification has become an interesting alternative to the input-output one, because of the possibility to use freely available ambient excitations (e.g. traffic, wind) and to accurately measure the responses. At the same time it entails restrictive assumptions, such as the use of a stochastic stationary process as input. The time domain subspace methods are widely used in the identification of linear time-invariant (LTI) systems. Among them, the Data-Driven Stochastic Subspace Identification (Data-SSI) is particularly attractive because of the reduced computational effort due to direct use of the output data, collected in a Hankel matrix. One of the first steps in Data-SSI is the partition of this matrix into two submatrices which have either the same number of block rows (symmetric partitions) or different (asymmetric partitions). The effects of asymmetric partitions on the identification are not well investigated in Literature. The objective of this paper is to evaluate the performance of Data-SSI in the identification of the modal parameters of a steel frame model tested on shaking table with stationary and non-stationary excitations using both symmetric and asymmetric partitions. The results are compared with those from an input-output subspace technique.