Date of Completion

1-16-2017

Embargo Period

7-15-2017

Keywords

Structural Health Monitoring, Structural Damage Detection, Structural Control, MR Damper, Bridge Monitoring, Machine Learning, Big Data Analytics, System Identification, Extend Kalman Filter, Statistical Process Control

Major Advisor

Shinae Jang

Associate Advisor

Richard Christenson

Associate Advisor

Jeongho Kim

Field of Study

Civil Engineering

Degree

Doctor of Philosophy

Open Access

Open Access

Abstract

The overall goal of the research in this dissertation is to develop efficient and accurate algorithms to detect damages in real time for civil structures using large scale of monitoring data. First, big data analysis is performed using one-year monitoring weigh-in-motion data collected from an in-service highway bridge located in Meriden, Connecticut. In order to quickly evaluate the structural condition from large scale of weight-in-motion data, two structural reliability-based methods were developed based on yield strength limit state and fatigue limit state. Then, in order to estimate structural parameter from structural dynamic response data, extended Kalman filter is used to estimate stiffness for a three-story building with MR damper. All the above methodology will be further developed for real-time damage detection for building structure as well as bridge structures. For building structures, a novel real-time structural damage detection method is developed for building structures that can be represented using structural dynamic models by integrating extended Kalman filter and dynamic statistical process control. The numerical validation is performed on a three-story linear building structure and a two-story nonlinear hysteric building structure, considering different damage scenarios during earthquake excitation. The simulation results demonstrate high detection accuracy rate and light computational costs of this method. The EKF-based method is further developed for full-scale implementation on bridge structures. An extended Kalman filter-trained artificial neural network (EKFNN) method is developed to eliminate the temperature effects and detect damage for a long-term monitored highway bridge. Numerical testing results show that the temperature-induced changes in natural frequencies have been considered prior to the establishment of the bridge damage warning thresholds, and the simulated damages have been successfully captured in real time. The research work in this thesis can provide engineers and scientists a thorough understanding of how to process and analyze big data collected from structural health monitoring systems for real-time structural damage detection purpose. This study will also have practical importance for infrastructure owners (e.g., Department of Transportation, building owners, etc.) and first responders (e.g., policemen, fire fighters, rescuers, etc.) to make rapid decisions after structural damage events.

COinS