Date of Award

5-1-2026

Degree Name

Master of Science

Department

Civil Engineering

First Advisor

Tezcan, Jale

Abstract

Modal identification plays an important role in vibration-based structural assessment. In civil engineering structures, input excitation data are often difficult to obtain, which makes output-only modal identification methods practically important. Independent Component Analysis (ICA) performs well for lightly damped systems, but its accuracy decreases as damping increases because rapid response decay weakens the statistical independence needed for effective mode separation. To address this limitation, a modified framework combining the Random Decrement Technique (RDT), Inverse Damping Transfer (IDT), and ICA was previously proposed. However, previous studies mainly focused on noise effects and were limited to systems with fewer modes and well-separated modal frequencies. In this study, the performance of the modified ICA framework is evaluated using numerically simulated responses from a five-degree-of-freedom structural system. Both free-vibration and ambient-vibration cases are considered under varying damping levels. The study focuses on closely spaced modal conditions, the effect of ambient response duration, and the robustness of the method relative to classical ICA. A comparative study of the ICA-only, RDT+ICA, and RDT+IDT+ICA methods is also carried out to investigate the individual effect of RDT preprocessing on ICA and, subsequently, the combined effect of RDT and IDT on the performance of ICA-based identification. Results show that, even in the presence of closely spaced modes and at higher damping levels, the modified ICA framework yields more reliable estimates of modal parameters, including natural frequencies, mode shapes, and damping ratios, than classical ICA. The study further highlights the influence of response duration on identification performance, showing that longer response records improve the accuracy and consistency of the identified modal parameters. Overall, the combined RDT+IDT+ICA framework provides a more robust and reliable approach for output-only modal identification of multi-degree-of-freedom systems, with consistent performance observed under free-vibration conditions and across repeated runs using different random initializations.

Share

COinS
 

Access

This thesis is only available for download to the SIUC community. Current SIUC affiliates may also access this paper off campus by searching Dissertations & Theses @ Southern Illinois University Carbondale from ProQuest. Others should contact the interlibrary loan department of your local library or contact ProQuest's Dissertation Express service.