Structural health monitoring (SHM) is an emerging interdisciplinary research field that aims at providing an intellectual framework to ensure the safety and performance of civil structures. By incorporating sensing technologies and signal processing techniques, vibration-based SHM extracts information of monitored structures from vibration measurements to assess their operational conditions, e.g. loading and response time histories, and damage conditions, e.g. locations and severity of damages. With the health information, we can prevent structural failures without early warnings and improve structural designs in the future.

With the advance in computational power, online monitoring is able to provide real-time information about the health status of structures. It also has poten...[ Read more ]

Structural health monitoring (SHM) is an emerging interdisciplinary research field that aims at providing an intellectual framework to ensure the safety and performance of civil structures. By incorporating sensing technologies and signal processing techniques, vibration-based SHM extracts information of monitored structures from vibration measurements to assess their operational conditions, e.g. loading and response time histories, and damage conditions, e.g. locations and severity of damages. With the health information, we can prevent structural failures without early warnings and improve structural designs in the future.

With the advance in computational power, online monitoring is able to provide real-time information about the health status of structures. It also has potential applications in structural control. In this thesis, online structural health monitoring using Kalman-filter-based methods for state estimation and force identification are presented.

Online joint input-state estimation is a useful tool for reconstructing the dynamical response and time-varying input time histories of a structural system in real time. Variants of Kalman filter (KF) have been developed to accomplish this task. A case study is performed to compare the performance of these new filters when subjected to a restriction on the sensor type.

Taking into considering a limited number of sensors, an optimal sensor placement (OSP) algorithm provides an optimal design of a sensor network in order to minimize the reconstruction errors before the installation of sensors. A new OSP algorithm is proposed based upon a recently developed joint input-state estimation method, the dual Kalman filter. The new algorithm is verified numerically.

A Bayesian framework for performing online noise parameter updating is presented in order to improve the estimation quality and uncertainty quantification ability of KF-based methods by automatically accounting for the unknown modeling errors.