Structural health monitoring is a process for identifying damage in civil infrastructures using sensing system. It has been increasingly employed due to advances in sensing technologies and data analytic using machine learning. A common problem within this scenario is that limited data of real structural faults are available. Therefore, unsupervised and novelty detection machine learning methods must be employed. This work presents a clustering based approach to group substructures or joints with similar behaviour on bridge and then detect abnormal or damaged ones, as part of efforts in applying structural health monitoring to the Sydney Harbour Bridge, one of iconic structures in Australia. The approach is a combination of feature extraction, a nearest neighbor based outlier removal, followed by a clustering approach over both vibration events and joints representatives. Vibration signals caused by passing vehicles from different joints are then classified and damaged joints can be detected and located. The validity of the approach was demonstrated using real data collected from the Sydney Harbour Bridge. The clustering results showed correlations among similarly located joints in different bridge zones. Moreover, it also helped to detect a damaged joint and a joint with a faulty instrumented sensor, and thus demonstrated the feasibility of the proposed clustering based approach to complement existing damage detection strategies.

Automated techniques for analyzing the dynamic behavior of full-scale civil structures are becoming increasingly important for continuous structural health-monitoring applications. This paper describes an experimental study aimed at the identification of modal parameters of a full-scale cable-stayed bridge from the collected output-only vibration data without the need for any user interactions. The work focuses on the development of an automated and robust operational modal analysis (OMA) algorithm, using a multistage clustering approach. The main contribution of the work is to discuss a comprehensive case study to demonstrate the reliability of a novel criterion aimed at defining the hierarchical clustering threshold to enable the accurate identification of a complete set of modal parameters. The proposed algorithm is demonstrated to work with any parametric system identification algorithm that uses the system order n as the sole parameter. In particular, the results from the covariance-driven stochastic subspace identification (SSI-Cov) methods are presented.

A large-scale cable-stayed bridge in the state of New South Wales, Australia, has been extensively instrumented with an array of accelerometer, strain gauge, and environmental sensors. The real-time continuous response of the bridge has been collected since July 2016. This study aims at condition assessment of this bridge by investigating three aspects of structural health monitoring including damage detection, damage localization, and damage severity assessment. A novel data analysis algorithm based on incremental multi-way data analysis is proposed to analyze the dynamic response of the bridge. This method applies incremental tensor analysis for data fusion and feature extraction, and further uses one-class support vector machine on this feature to detect anomalies. A total of 15 different damage scenarios were investigated; damage was physically simulated by locating stationary vehicles with different masses at various locations along the span of the bridge to change the condition of the bridge. The effect of damage on the fundamental frequency of the bridge was investigated and a maximum change of 4.4% between the intact and damage states was observed which corresponds to a small severity damage. Our extensive investigations illustrate that the proposed technique can provide reliable characterization of damage in this cable-stayed bridge in terms of detection, localization and assessment. The contribution of the work is threefold; first, an extensive structural health monitoring system was deployed on a cable-stayed bridge in operation; second, an incremental tensor analysis was proposed to analyze time series responses from multiple sensors for online damage identification; and finally, the robustness of the proposed method was validated using extensive field test data by considering various damage scenarios in the presence of environmental variabilities.