STUDY ON ACOUSTIC EMISSION DETECTION FOR PIPELINE LEAKAGE BASED ON EMD SIGNAL ANALYSIS METHOD

A signal analysis method based on empirical mode decomposition(EMD) is proposed aiming at extracting non-stationary characteristic of an acoustic emission (AE) signal result from pipeline leakage. The detected leakage signal can be decomposed into a sum of finite intrinsic mode functions (IMF) by EMD, some of IMFs embracing typical AE characteristic can be selected to be reconstructed. Thus, noise interference can be eliminated. The reconstructed signals are analyzed and their cross-correlation parameter is calculated. The locating accuracy of the leaking point calculated with the reconstructed signals is greatly increased. It is demonstrated that Hilbert-Huang transformation is an effective tool for demonstrating the unstable characteristic of acoustical signals and feature extraction of pipeline leak signals.