Document Type



Doctor of Philosophy


Structural Engineering

First Adviser

Pakzad, Shamim N.

Other advisers/committee members

Wilson, John L.; Sause, Richard; Baird, Henry S.; Venkitasubramaniam, Parvathinathan


The goal of this dissertation is to advance the state-of-art of data-driven structural monitoring, which is a promising way to automate the maintenance process of civil structures, thus benefiting their life-cycle management in both the economy and performance aspect. The availability of affordable electronic data acquisition systems enabled continuous monitoring of structures, and effective data compression algorithms are needed for structural state characterization/damage detection from the collected signals. Ideally, the data processing algorithms would provide information for all four levels of damage detection proposed by Rytter: 1) damage existence; 2) damage location; 3) damage severity; 4) remaining service life prediction. Damage detection in real-world structures is a complex problem because of the various possible forms of damage that can occur and the influences of operational/environmental variations on the observations/measurements made. The focus of this study will be the first three stages of damage detection using vibration measurements, which are commonly measured for structural health monitoring purposes. For damage existence identification time series analysis on single channel response will be used (Part I of this document), while higher-level damage detection is attempted by using multi-input-single-output subsystem modeling. More detailed outlines of both subjects can be found in their respective introduction sections.The specific contributions of this study are in the following technical areas: exploring the capabilities of different SHM vibration sensors, proposing and testing new indicators/thresholds for damage identification, cross-comparing the proposed indicators/thresholds with existing ones through applications to various civil structures, and developing theoretical models regarding the validity/sensitivity/robustness of several damage features. All these efforts are for the search and development of optimal damage detection method in a certain application.Farrar et al. proposed a statistical pattern recognition (SPR) paradigm for vibration-based structural health monitoring, which quite well generalized most damage detection procedures. The paradigm contains 4 steps: 1) Operational evaluation; 2) Data acquisition and cleansing; 3) feature selection and data compression, and 4) statistical model development. Research conducted for this dissertation centers on the last three stages of this paradigm, with most of the technical contributions in the last two categories.