Acoustic emission technology is mostly used in corrosion detection of the atmospheric vertical storage tank bottom, but the evaluation results are always affected by the complex sound sources. In this paper, wavelet neural network is used to identify the acoustic emission signals from different types of tanks. Using wavelet transform and threshold denoising to denoise the detection signals, after wavelet packet decomposition, each node’s energy distribution and the feature vectors of extracted corrosion signals of the tank floor are selected as the input. At last, the compact-type wavelet neural network is chosen to recognize different AE signals. The result of magnetic flux leakage test proves that this method can improve acoustic emission signal analysis precision and achieve the accurate corrosion evaluation based on AE technology.

This content is only available via PDF.
You do not currently have access to this content.