Detection of damage in civil engineering structures by PCA on enviromental vibration data

The dynamic behavior of civil engineering structures are usually studied by means of ambient vibration observations and their performance is analyzed by Peak Picking and/or Operational Modal Analysis methods. This paper reports the first results of a statistical multivariate approach, specifically Principal Component Analysis, to detect a suspected structural damage on a sicilian highway bridge. Furthermore, the damage simulated in a simple structural model made it possible to understand the characteristics of the method consisting in comparing the observed data on an undamaged structure with those coming from a damaged one.