System identification and model updating of highway bridges using ambient vibration tests

Abstract

Vibration-based structural health monitoring (SHM) has received significant attention in the past. Due to the existence of some defect of implementation, the measured response of a structure and the response from its finite element model may not match. There are a number of methods available for updating the Finite Element (FE) model of a structure such that the response calculated from the model agrees with field measurements, and identifying the system parameters like stiffness and mass based on dynamic response of the structure. These methods are categorized into physics based and data driven. In this study, the FE models of a 16-span Pre-Stress Concrete Box (PSCB) girder bridge with the total length of 780 m, a 3-span Void Slab Bridge with the total length is of 65 m, and a Steel Box (STB) with 380 m length and 12 m width, with 8 spans of equal length of 47.5 m bridge are constructed and updated using the measured vibration data. The objective of this study is to identify the system properties of the bridges using physics-based and data-driven methods and update the corresponding models using the data from ambient vibration tests and determine the efficacy of each method. A well-known and effective physics-based method, the matrix update method, is used for correlating the models by solving the relevant inverse problem through constrained optimization. In data-driven methods, the Neural Network and Genetic Algorithms are applied to find the correlations between the structural frequencies and changes in the sectional properties of the bridge segments. The outputs of these models are compared with certain target frequencies based on the measured data in order to adjust the section properties of the bridge elements. It is found that while the physics-based method has a better performance than the data-driven model in identifying the modal properties, the physics-based model is difficult to implement and there is a need for developing a hybrid method to achieve a better result.