ABSTRACT
During the past several decades, a significant amount of research has been conducted in the area of damage detection via changes in the modal response of a structure. Also, research for damage detection in structures using statistical analysis is very active. Two classes of statistical models that have been presented are unsupervised learning and supervised learning. Unsupervised learning approach is limited to level 1 damage detection (i.e. alarming the occurrence of damage) or level 2 damage detection (i.e. identifying the location of damage), but supervised learning approach can be taken to move forward to higher level damage identification for classifying, locating, and quantifying the severity of damage.
