Acoustic emission detection and classification using wavelet-based power-law detector

Acoustic Emission (AE) technology is capable of continuously monitoring micro-structural changes in materials and structures. To discriminate true AE signals from environmental noises is essential for AE technology. In this paper, an improved power-law detector with discrete wavelet packet transform (DWPT) and best basis selection (BBS) algorithms was developed to detect and classify transient AE signals. DWPT was first used to decompose an acoustic signal into a set of orthogonal wavelet packets. Then BBS for the power-law detector was determined based on the prior knowledge of the AE signals and noises. An experimental setup was built to test the performance of DWPT-based power-law detector. Four types of acoustic signals (including real AE and simulated acoustic events) were produced in lab conditions. The test results showed that the detection rate was close to 100%, while the false positive rate was less than 2%.