Successive structural monitoring using wavelet packet signature

This paper proposed a covariance-driven wavelet packet feature extraction method to monitor structural health monitoring during normal operating conditions. The unique feature of this method is it only requires output response measurements at one location of the structure and thus it is very suitable for practical application. This new method is based on the combination of wavelet packet feature extraction technique proposed earlier by the authors and a covariance-driven pre-processing procedure to deal with the output-only problems. Two damage indicators are formulated based on the extracted wavelet packet component energies. Statistical process control technique is then adopted for on-line monitoring of the damage,indices. The efficiency, accuracy and robustness of the proposed method are illustrated by a numerical study on the ASCE benchmark frame model. The results show the proposed method has a great potential for the online monitoring of structural health condition. Different levels of simulated measurement noises are added into the response records to check the noise sensitivity of the proposed method. The results are also very excited: the proposed method seems to be immune from 0dB noise, which represents the noise has the same magnitude as the signal in the measurement