A new data-fusion method is proposed for the damage detection of anisotropi
c composite materials. Based on signal processing theory, we combine wavele
t packets, which can decompose signals into a tiling of plane of the time f
requency and the feature information in different frequency bands, with aut
oregression spectrum analysis to extract features and recognize the charact
eristic signals sampled at experiments of damage detection of composites by
vibration. These features are fed into the wavelet neural network as the i
nput patterns for training and classifying. Analysis on the signals obtaine
d in the damage detection experiment of composites demonstrates the effecti
veness of the proposed method.