The improvement and application of acoustic emission inspection algorithm for metal vessel

A method of acoustic emission defect inspection based on wavelet packet analysis and BPNN (BP neural network)is introduced. The method of wavelet packet based on sections and energy-moment feature is used to replace the traditional “wavelet packet-energy” to pick-up characteristics of AE signals. The efficiency of this method is validated by experiment of metal vessel defect diagnosis. The result shows that compared with ordinary way, the method of feature extraction based on wavelet packet of sections and energy-moment feature, can make better use of the major band of defect signals and the wavelet´s time-frequency information, and reduce the complexity of system and increase the identification rate.