Wavelet packet method of damage feature extraction in bridge health monitoring

For structural damage identification in bridge health monitoring,feature extraction of damage pattern classification was proposed and analyzed,and two feature exaction methods based on wavelet packet(WP) node energy were presented.To study the difference of WP coefficient node energy and WP signal component node energy,a numerical simulation of continuous beam impacted by random loads was conducted and acceleration signals of different structural condition were obtained.WP was applied to extract two kinds of node energy feature from the acceleration signals.Comparison of the two kinds of feature index showed that there is little difference in their sensitivity.Due to the very good computation efficiency of the WP coefficient,it was concluded that feature index of the WP coefficient node energy can satisfactorily meet the needs of bridge health monitoring.