Feature Extraction of Pipeline Crack Defect Signals with MMM Testing Based on Wavelet Packet Frequency Bands Energy

Abstract

In order to solve the problem of defect criterion that can not effectively identify the Stress concentration region and crack defect, wavelet packet transformation was used to multiscale wavelet analysis of Metal Magnetic Memory (MMM) signals. A new signal inspection technology was presented based on energy increment feature and wavelet packet frequency bands, which can greatly perfect the criterion. The Daubechies wavelet was used as a wavelet packet function with the series of three, and wavelet packet frequency bands energy method was used to analyze MMM signals. Comparing the frequency bands energy increment of the Stress concentration region and crack defect, the threshold was established that would realize the accurate testing of in-service pipeline crack defect. After de-noising of MMM signals, the power feature extraction was completed by virtue of experiment. While compared with the testing result of Flux Leakage Magnetic (FLM) method, the new technology can effectively identify pipeline crack defects. The theoretical basis was provided for pipeline crack defect identify with MMM testing.