Irregular Continuum Structures Damage Detection based on Wavelet Transform and Neural Network

This paper presents a method for detecting damage in irregular 2D and 3D continuum structures based on combination of wavelet with neural network. The method proposed here only requires the responses (displacements, stresses) of the damaged structures, while most damage detection methods need the structural responses before and after damage. First, the structural responses related to the damaged state are determined at the finite element points having irregular distances. Secondly, the Multiple-Layer Perceptron Neural Network (MLPNN) is used to estimate the responses at points having equal distances by those previously obtained by the finite element. Then, the extended responses are analyzed with the 2D and 3D wavelet transform in order to locate damaged zones. It is shown that detailed matrix coefficients of 2D and 3D wavelet transform can identify the damaged zone of the structure by perturbation in the damaged area. In order to assess the performance of the proposed method, some numerical examples are considered. The results show the high efficiency of the method for damage localization of the structure.