Damage Detection in Lightweight Structures Using Artificial Intelligence Techniques

Abstract

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.