Tracking of Structural Dynamic Characteristics Using Recursive Stochastic Subspace Identification and Instrumental Variable Technique

Identifying dynamic characteristics for structures under ambient condition is a very important task for the design verification, model updating, health monitoring, and damage detection of these structures. Among many techniques developed for this purpose, the stochastic subspace identification (SSI) techniques are a robust time-domain method that can simultaneously process large numbers of inputs and outputs. They can potentially extract the most information from the measurement. The SSI method assumes that both the excitation and the measurement noise are stationary white-noise processes and can lead to some errors when this assumption is not met. In this paper, a recursive SSI method using an instrumental variable (IV) technique is proposed for online step-by-step tracking of time-varying modal parameters for a structure under nonwhite but finitely correlated excitation and measurement noise. The IV technique is used to resolve possible estimation bias attributable to the violation of white noise assumption for the SSI method. A bi-iteration subspace tracker is also adopted to make the technique efficient and suitable for online application. The proposed technique is demonstrated on a single-degree-of-freedom structure and the ASCE benchmark frame structure. Results show that the technique can track the time-varying characteristics of structures under nonwhite but finitely correlated excitation and measurement noise.