Intelligent identification and localization of distributed optical fiber pipeline leakage based on VMD-WD and SSA-BP...

To address challenges such as weak pipeline leakage signal characteristics and the difficulty of leakage identification, this study explores the application of distributed acoustic sensing (DAS) technology in pipeline leakage monitoring. Specifically, leakage signal denoising, condition identification, and spatial localization were investigated, and an integrated pipeline leakage monitoring method comprising signal enhancement, condition identification, and spatial localization was developed. First, variational mode decomposition (VMD)-based denoising was combined with wavelet decomposition (WD)-based mode screening to suppress background noise. Second, a backpropagation (BP) neural network optimized by the sparrow search algorithm (SSA) was developed for leakage identification. Finally, accurate leakage localization was achieved using the high-to-low-frequency energy ratio as the primary criterion. Experimental results showed that with the leak location fixed and tested under three different flow-rate conditions, the maximum absolute positioning error was estimated at 0.4 m, corresponding to a positioning accuracy of 98.1%. Under the experimental setup and test conditions described in this paper, the proposed method demonstrates effective leak localization capabilities; it successfully enhances leak characteristics, improves the accuracy and stability of identifying various leak states, and achieves precise leak point localization, thereby offering significant value for engineering applications.