Speaker
Description
Damage localization is a crucial process of a data-based structural health monitoring strategy. This paper presents a study for identifying damage lo-cations by proposing two statistical dam-age indices based on reliability index for normally distributed independent random data. Auto-Regressive (AR) model is fitted to the measured vibration time-domain response in an effort to extract the model residuals as the randomly damage-sensitive fea-tures. The proposed indices are based on computing the direct and relative differences between the reliability indices of the AR model residuals of the undamaged and damaged structures. Statistical damage localization based on the proposed damage indices lies in the fact that their largest quantities are indicative of dam-age locations. A series experimental acceleration time series data from a three-story laboratory frame are applied to demonstrate the capability and accuracy of the proposed methods. Results will show that both of the damage indices are capable of identifying the location of damage even when operational and environmental variability are available.
