On the use of parallel full-order state observers for damage detection in mechanical systems

According to the literature, structural health monitoring techniques based on state observers usually evaluate the differences between measured and estimated states to detect a damage. The practical application of this strategy is an unfeasible solution if the monitoring system needs to detect damag...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Journal of the Brazilian Society of Mechanical Sciences and Engineering 2021-12, Vol.43 (12), Article 556
Hauptverfasser: Mattei, Rafael D., de Melo, Gilberto P., Bueno, Douglas D.
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:According to the literature, structural health monitoring techniques based on state observers usually evaluate the differences between measured and estimated states to detect a damage. The practical application of this strategy is an unfeasible solution if the monitoring system needs to detect damages in a region of the structure where sensors cannot be placed. For these cases, a common alternative is to use sensors at nearby regions, but the accuracy of detection can be reduced. In this context, the present article introduces a new approach for damage detection in mechanical systems using the relative difference obtained from the estimations of a state measured by two different sensors. In this method, the estimated state is obtained by sensors in different positions from the one that needs to be monitored. To do so, parallel full-order state observers are employed and an optimal gain is computed. An index is proposed to detect the damages, which is time-dependent and computed using the responses of the systems. Numerical simulations are carried out considering two different mechanical systems. Linear and nonlinear damages are considered, and the results demonstrate that the proposed approach can detect damages accurately.
ISSN:1678-5878
1806-3691
DOI:10.1007/s40430-021-03270-4