Wavelet Approach to Damage Detection of Mechanical Systems and Structures

Abstract

The damage detection methods of mechanical and civil structures have been drawing much interest from various fields. Many techniques to detect damage are based on the examination of the system response signals. Crack-like damages may contribute to the response signal the edges (also called slopes), i.e. the localized sharp transitions of signal values. A powerful tool to characterize such local feature signals is the continuous wavelet transform (CWT). Despite its name, the CWT can be calculated on discrete data and requires enormous computational cost. However, in many cases, the crack-like changes in the signal can be detected and localized using the discrete dyadic wavelet transform (DDWT) that has a fast transform algorithm. This paper presents application of the DDWT to detect and localize the response signal features (slopes) due to cracks in mechanical and civil structures. The wavelets used to calculate the DDWT are cubic box splines. The numerical results for a response signal simulating cracks in structures are presented in this article.