In previous works by the authors, it was shown the advantages of applying an active piezoelectric system in combination with Principal Component Analysis and Neural Networks as methodology for damage detection in structures. An active piezoelectric system considers the advantage of using piezoelectric transducers (PZT's) as actuator as well as sensor. In each phase of the diagnosis procedure, one PZT is used as actuator (a known electrical signal is applied) and the others are used as sensors (collecting the wave propagated through the structure at di erent points). An initial baseline model for undamaged structure is built applying Principal Component Analysis (PCA) to the data collected by several experiments. Current structure (damaged or not) is subjected to the same experiments, and the collected data are projected in the PCA model. In this paper, two indices are used to detect damages; these indices are calculated from the information obtained from the projection of the experiments in the PCA model (baseline). Besides the localization is performed using ve di erent methods. These methods are based on the contribution of each sensor to each index, in this way, according to these contributions, the damage can be localized. The combination of all indices and all contributions (a total of 2 x 5) are analyzed and compared. To validate the approach, the methods are applied to an aluminum plate which is instrumented with several PZT's.

Peer Reviewed