Modeling of Environmental Effects for Vibration-Based SHM Using Recursive Stochastic Subspace Identification Analysis

Abstract

This paper deals with the problem of a bridge structure identification using output-only vibration measurements under changing environmental conditions. Two key issues of a real-life monitoring system are addressed through analysis. The first issue is the identification of structural dynamic characteristics directly from measurements under operating conditions. The covariance-driven recursive stochastic subspace identification (RSSI-COV) algorithm is applied to extract the system dynamic characteristics. The second issue is to distinguish the system dynamic features caused by abnormality from those caused by environmental and operational variations, such as temperature, and traffic loading. In this study a solution is proposed to model and remove the uncertainty due to environmental effects from the identified system dynamic characteristics from on-going measurements. Nonlinear principal component analysis incorporated with AANN is employed to distinguish the dynamic feature changes caused by abnormality from those caused by environmental and operational variation (i.e. ambient temperature and traffic loadings). Finally, field experiment of a bridge is conducted. The variation of the identified system natural frequencies was discussed by using the proposed method.