Structural Damage Detection Based On The Correlation Analysis Between The Wavelet Packet Component Energies

Yan GR,Duan ZD,Ou JP

2004-01-01

Abstract:Since wavelet packet transform (WPT) performs a complete decomposition not only in low frequency band but also in high frequency band, it provides a promising tool for structural damage detection. Most of damage detection approaches previously developed using WPT employ an artificial neural network with the wavelet packet energies in each frequency band as input vector. In this paper, a novel damage detection method based on the correlation analysis between the wavelet packet component energies (WPCE) extracted from the analytical data and measured data was presented. For the case of ambient vibration, the WPCE extracted directly from the responses can not be used in this study, because the WPCE depend on the applied force and the ambient excitations of a structure in the pre and post damage states might be quite different. Therefore, the WPCE were extracted from the cross-correlation functions between two measured responses instead of responses themselves in this investigation. The proposed WPCE eliminates the influence of external loads. Sensitivity analysis of the WPCE to the damage extent of all elements was preformed to simulate all kinds of damage scenarios for complex structures. This method can be applied to the location of multiple damage sites and can be used to determine the damage extent. The benchmark steel-frame scale structure presented by IASC-ASCE structural Health Monitoring Task Group was implemented to illustrate the versatility of the presented method.