Search

Reliable and efficient damage detection is critical for the use of lightweight materials in the mechanical and aerospace industries. Within the context of Non-Destructive Testing (NDT), vibration-based tests have been applied for many decades to inspect components without damaging or debilitating their use. For posterior fault recognition, Artificial Intelligence techniques have achieved high success for a number of structural applications. In this work Testing, Simulation and Artificial Intelligence have been combined in order to develop a defect detection procedure. The use of an Optomet Scanning Laser Doppler Vibrometer (SLDV) for such tests provides an interesting solution to measure the vibration velocities on the structure surface. The algorithm for identifying the defects is based on the Local Defect Resonance (LDR) concept, which looks to the high frequency vibrations to get a localized resonant activation of the defect. Artificial Intelligence (AI) techniques were implemented with the aim of creating an automatic procedure based on features extraction for damage detection. Wavelet transformation and modal analysis were used to provide inputs to the AI techniques. In order to better understand the limitation in terms of defect detection, damaged plates were modelled and simulated in order to perform a sensitivity analysis. Finally, an overall comparative overview of different algorithms results was also obtained. [ABSTRACT FROM AUTHOR]

The objective of this paper is the investigation of the capability of operational modal analysis approaches to deal with time-varying system in the low-frequency domain. Specifically, the problem of the identification of the dynamic properties of a launch vehicle, working under actual operative conditions, is studied. Two OMA methods are considered: the frequency-domain decomposition and the Hilbert transform method. It is demonstrated that both OMA approaches allow the time-tracking of modal parameters, namely, natural frequencies, damping ratios, and mode shapes, from the response accelerations only recorded during actual flight tests of a launcher characterized by a large mass variation due to fuel burning typical of the first phase of the flight. [ABSTRACT FROM AUTHOR]

Output-only analysis considers the system excited either by operative or by natural forces, in both cases the input loading which excites the structure is considered, at least in a limited frequency band, as a white noise. Because it is not possible to find directly the spectra, it is necessary to pass through the correlation functions so as to apply the Wiener-Khintchine theorem in order to find the power spectral densities (PSDs). In the past, the modal parameters have been essentially derived by approaches that manipulated the functions mentioned above. In this paper, the modal parameters will be directly estimated from the PSDs, in the frequency domain, and from the correlation functions in the time domain. A particular attention is devoted to the problems regarding the damping ratio overestimation. This effect due to the limited time window in the correlation function estimate is highly present in the low-frequency modes and could bring to large estimation errors. Experimental examples, carried out both on cantilever beams and on an helicopter blade, are presented. They show the problems related to the techniques using the PSDs, or the correlation functions and the possibility to overcome the overestimation of the damping factors due to the triangular window, also known as Bartlett window. [ABSTRACT FROM AUTHOR]

The article presents an analysis of experimental and numerical procedures that analyze an AB 204 helicopter rotor blade's structural properties developed by L. Balis Crema and G. Coppotelli. It describes the development of the dynamic behavior of the blade which is noted to be based on the frequency response-based correlation functions sensitivity to the design parameter changes. The finite element validation of the stiffness distribution of the helicopter rotor blade is evaluated.