MULTI-COMPONENT DECONVOLUTION INTERFEROMETRY FOR DETECTING AND LOCALIZING DAMAGE: A NUMERICAL STUDY

Author

Sujeet Gurung, Southern Illinois University CarbondaleFollow

Date of Award

8-1-2023

Degree Name

Master of Science

Department

Civil Engineering

First Advisor

Sen, Debarshi

Abstract

Damage identification in a structure is necessary because it can provide important information about the current condition of a structure and prevent human and financial losses. Traditionally, finite element (FE) models have been used for damage identification in structures. However, these FE models are limited by modeling assumptions and computational efforts associated with them. These limitations favor the data-based approaches and deconvolution interferometry is one of them which can extract impulse response functions (IRFs) from a structural system. These extracted IRFs can be utilized as features for damage detection. The traditional form (single component) is based on the assumption that direction of motion is independent of motions in other degrees of freedom which may not be true in many real-field scenarios. Multi-component deconvolution interferometry (MDI) addresses this assumption of interferometry based on single component and has been effectively used in dynamic characterization and response prediction for more accurate results than the single one. This thesis studies the possibility of detecting and localizing damage in a structure by using MDI approach. A numerical study is performed on a 3-D model of a five story single bay structure subjected to ambient vibrations both pristine and damaged conditions. Correlation coefficients and l_2-norm ratios are used to detect damage. Principal component analysis of the extracted features demonstrates promise on the ability to localize damage in structure in the presence of measurement noise.