Feature extraction of acoustic emission signals for cable damage based on wavelet packet analysis

The simulated feature extraction of acoustic emission signals for the damage of cables was presented on the basis of the combination of explicit finite element method and wavelet packet analysis by using the program of ANSYS/LS-DYNA. The feature extraction was discussed in the aspects of decomposition level, number of the selected feature bands and noise robustness of the feature parameters. The results show that: the wavelet packet energy spectrum can reflect the feature of the signals finely when proper decomposition level is chosen; only a minority of the feature bands can make the wavelet packet energy spectrum satisfactorily reflect the feature information of acoustic emission signals; the feature parameters established based on wavelet packet energy spectrum are of preferable damage sensitivity and noise robustness and the discrimination of various damage types can be realized in spite of the existence of high level of noise.