Natural gas pipeline leak aperture identification and location based on local mean decomposition analysis

When a gas pipeline leaks, the leak aperture and its position cannot be easily identified. This paper proposed a leak aperture recognition and location method based on root mean square (RMS) entropy of local mean deposition (LMD) and Wigner-Ville time-frequency analysis. Firstly, wavelet packet energy analysis is employed to remove noise and extract the primary energy leak bands to reconstruct. The reconstructed signal is then decomposed using LMD, and the RMS entropy of PF components is calculated. The RMS entropies of multiple groups are used to build the feature vector, and it is then input into support vector machines (SVMs) to achieve the aperture recognition. An adaptive method based on mutual information is proposed to select the sensitive product function (PF) components, and the time-frequency parameters of the sensitive PF components are then calculated using Wigner-Ville distribution (WVD) method. The time delays are obtained by analyzing the time-frequency parameters. Finally the leak location was achieved by combining time delay with leak signal propagation velocity. The experimental results show that the proposed method can effectively identify different leak apertures, and the leak location accuracy is better than that of the direct cross-correlation method.