Development of an indirect bridge health monitoring approach using moving sensors

Publication Type:
Thesis
Issue Date:
2019
Full metadata record
The inevitable deterioration and damage of bridge infrastructures due to repeated and excess traffic loading, environmental erosion and ageing are of great concern worldwide. Bridge structural health monitoring (SHM) is critical to obtain structural health information and early warning for potential damage. Most of the current SHM strategies measure vibration responses from sensors installed at different locations on the bridge. This direct approach poses several challenges, such as the high cost of the installation and maintenance of sensors, the need for extensive data processing and the insufficient spatial information. To seek a more economical and flexible way to monitor bridges, an indirect approach that measures responses of a passing vehicle has recently drawn great attention. This strategy involves the use of instrumented vehicles as a moving sensory system to capture bridge dynamic information via vehicle-bridge interaction (VBI). Sensors are installed on the vehicle axles or body. However, the responses from sensors on a moving vehicle are nonstationary, noisy and significantly affected by the surface roughness of the bridge. Therefore, most of the classical output-only system identification approaches based on the assumption of white noise excitation may fail to extract accurate structural dynamic properties. This research aims to establish a framework for bridge SHM using vehicle-based mobile sensory systems. Indirect structural identification methods that consider the intrinsic nonstationary characteristics of VBI responses are proposed to extract the bridge dynamic parameters from vehicle acceleration responses. Firstly, a Short-time Stochastic Subspace Identification (STSSI) strategy was proposed to identify bridge modal frequencies and mode shapes. This method combines conventional SSI with a rescale procedure to estimate the bridge modal parameters using the responses of two instrumented vehicles. Secondly, based on the sequential implementation of singular spectrum analysis (SSA) and blind source separation (BSS), a method named drive-by blind modal identification with singular spectrum analysis (SSA-BSS) was proposed to extract the response components from a single set of vehicle vibration responses. The bridge frequencies can be identified from the obtained bridge related components. Numerical and experimental results clearly demonstrated the feasibility and effectiveness of the proposed methods for indirect identification of bridge modal frequencies and mode shapes. To gain insight on the time-dependent features of VBI system, a time-frequency (TF) analysis method called Synchroextracting transform (SET) was used to analyse the vehicle and bridge responses in TF domain. The instantaneous frequencies (IFs) of the system revealed the time-varying characteristics of the VBI system. Besides the indirect bridge modal identification, a two-step drive-by bridge damage detection strategy using vehicle axle responses was proposed. Dual Kalman filter (DKF) was applied to identify the interaction forces between vehicle and bridge. With the interaction forces, a sensitivity analysis was performed with regularization technique to identify the bridge damage. The proposed two-step damage detection method effectively identified the location and extent of the damages using vehicle axle responses, which demonstrated its great potential for drive-by bridge damage detection. Moreover, the SSA-BSS and the TF analysis strategy were successfully applied to analyse the responses from an in-situ VBI system. In summary, an indirect bridge SHM technique using vehicle-based moving sensing system was developed in this study. Bridge modal identification and damage detection were conducted successfully using vehicle responses. Results further demonstrated that it can be a convenient and cost-effective alternative or a promising complement to conventional bridge SHM.
Please use this identifier to cite or link to this item: