Abstract:

The wavelet neural network(WNN) is applied in many fields for its advantages of fast convergence，simple structure and computation since it has been presented. According to its strongly approaching ability to single dimensional signals, the Grey Relevancy Analysis theory is adapted to analyze the characteristic parameters and damage index of structural damage. The obtained relevancy degree of parameters is regarded as characteristic parameters of dynamic structure, and is transformed into weight value of singledimension input. Furthermore, the WNN model is presented with the Grey Relevancy Analysis. In order to improve the convergence ability and approaching effect, the study algorithm of WNN is ameliorated by local study policy. The simulation result indicates that the proposed method is superior to traditional Neural Network model.

Key words: computer network, wavelet neural network(WNN), damage detection, grey relevancy analysis, local study policy