An optimized VMD-wavelet denoising method for leakage detection in water supply networks from acoustic emission signals

Abstract

Leakage detection in water supply networks is critical for infrastructure maintenance, while traditional methods relying on listening devices are inefficient and time-consuming. Acoustic emission (AE) technology has emerged as a promising alternative due to its non-destructive nature and minimal environmental impact. However, its effectiveness is significantly hindered by environmental noise, which degrades the signal-to-noise ratio (SNR) and complicates leakage detection. To address this challenge, we propose an optimized VMD-wavelet denoising method tailored for AE-based leakage detection. Our approach introduces three key innovations: Adaptive VMD parameter optimization using the Northern Goshawk Optimization (NGO) algorithm, ensuring optimal mode decomposition; Correlation-based IMF selection, effectively filtering out irrelevant components to enhance signal clarity; and Improved wavelet threshold denoising, which refines high-frequency components to maximize noise suppression while preserving leakage-related features. Extensive experiments on simulated and real-world datasets demonstrate that our proposed method outperforms conventional approaches, increasing the SNR from 20.27 to 30.58 (approximately a 50% increase) and achieving a high average leakage detection accuracy of 94.63%. Our work contributes to the advancement of pipeline monitoring technologies, providing a more effective solution for maintaining real-world water supply networks.