Discussion of several “RSID” algorithms to linear time-varying system and system with abrupt change of joint failure

Abstract

Tracking modal parameters and estimating the structural current state from building seismic response measurement can provide useful information for building safety assessment, therefore, on-line or recursive identification technique needs to be developed for building seismic response monitoring. In this paper, three recursive subspace identification algorithms are developed. One is the recursive subspace identification with Bona-fide LQ renewing algorithm (RSI-BonaFide), one is recursive subspace identification based on matrix inversion lemma algorithm (RSI-Inversion) either with oblique projection technique or with orthogonal projection technique. A forgetting factor is introduced in the RSI-Inversion to emphasize the latest state of the time-varying system in this method. Two different sets of building seismic response data are used to verify the applicability of the proposed methods. One is the 3-story steel structure with abrupt change of story stiffness shaking table test in laboratory, the other is the seismic response analysis of a 9-story mid-story isolation building.