Citation

Candelieri A, Sormani R, Arosio G, Giordani I, Archetti F. Int. J. Reliab. Saf. 2013; 7(3): 216.

Copyright

(Copyright © 2013, Inderscince Publishers)

DOI

10.1504/IJRS.2013.057091

PMID

unavailable

Abstract

Online assessment of the structural health of aircrafts is crucial both in military and civilian settings. In this paper, Artificial Neural Networks (ANNs) are exploited to obtain a reliable system performing two tasks: diagnosis and prognosis. Diagnosis is devoted to (a) detect a crack, (b) identify the component of the panel involved (bay or stringer) and (c) estimate crack centre and size. Prognosis aims at estimating the evolution of the crack and the Remaining Useful Life (RUL). Training of the ANNs is performed on data sets built through finite elements simulation. Two different ANN hierarchies are presented for diagnosis. Crack evolution is performed for cracks on bay and stringer, separately. Two ANNs are used to estimate the parameters of a crack propagation model (NASGRO equation) for RUL prediction.

KEYWORDS: structural health monitoring; artificial neural networks; hierarchies of classifiers; helicopter fuselage; fuselage panels; aircraft structures; helicopters; reliability; fault diagnosis; crack detection; finite element analysis; FEA; simulation; bays; stringers; aluminium panels; crack propagation models; modelling; crack localisation.