Multi-point Leakage Localization in Natural Gas Pipelines Using Adaptive Wavelet Denoising and FastICA Blind Source Separation

During the long-term use of natural gas pipelines, leakage can easily occur due to factors such as corrosion and aging, which poses a threat to the surrounding environment and personnel safety. In this paper, the multi-point leakage location in natural gas pipelines under noisy environments is studied. Based on the diffusion characteristics of infrasound waves generated when a natural gas pipeline leaks, Fluent software is used to simulate the multi-point leakage condition in the pipeline and extract the acoustic signals. To realize the feature extraction of multi-point leakage source signals in a noisy background, a TDOA localization method combining adaptive wavelet threshold denoising and an improved FastICA (Fast Independent Component Analysis) blind source separation technique is proposed. The interference of noise on signal separation is suppressed through adaptive wavelet threshold denoising; the FastICA algorithm based on the maximum objective function of negative entropy is then applied to separate the independent leakage source signals. Combined with generalized cross-correlation (GCC) time delay estimation, the time delay between the observed signals and the leakage source signals is calculated. Finally, the leakage point positions are determined using a one-dimensional pipeline multi-point leakage model. Experimental results show that the independent characteristic signals of three leakage sources were successfully separated from four-dimensional observed signals containing three leakage points, and the positioning accuracy achieved using the generalized cross-correlation TDOA method was over 97%. The proposed method combining adaptive wavelet denoising and improved FastICA effectively addresses the challenges of signal separation and localization in pipeline multi-point leakage scenarios.