Two-step damage identification approach for a multi-layer concrete frame structure based on wavelet analysis and BP neural network

摘要

Wavelet analysis can easily detect the damage locations in a structure, but it is hard to determine the damage degree accurately. If the BP neural network is only applied to identify the damage, the training sample is too large to be impractical in some cases. To solve this problem, a two-step damage identification approach is then developed. In the first step, wavelet analysis is used to detect the damage locations. Then, the damage degree can be determined quantitively by using the BP neural network with a small size of training samples. A multi-layer concrete frame structure is investigated in this work. The result shows that the proposed approach is valid and practical.