变模式分解在供水管道漏点定位中的应用研究

针对供水管道中泄漏声信号的非平稳特征，提出了一种基于变模式分解（Variational Mode Decomposition,VMD）的漏点定位方法。算法首先利用VMD将传感器所采集的两路泄漏声信号按预设尺度自适应地分解为若干个固有模态函数，而后根据各模态函数与原始信号的相关系数选择分量重构信号，并达到消噪的目的。通过对消噪后的信号作互相关实现时延估计，最终实现对漏点的定位。实验结果表明：VMD分解方法可以在保留泄漏声信号本质特征的前提下，消除噪声对信号的干扰。较传统的EMD方法而言,其定位精度更高。

Research on leak location method of water delivery line based on Variational Mode Decomposition

A leak point location method based on Variational Mode Decomposition(VMD) is proposed to process the leakage acoustic signal in water supply pipeline with non stationary characteristics. Two leak acoustic signals collected by sensors are decomposed into several intrinsic mode functions according to the preset scale using VMD method first. Then in order to remove the noise, the signals are reconstructed by modes selected in accordance with the correlation coefficient of each mode function and the original signal. Finally the cross correlation of the filtered leakage acoustic signals is calculated to estimate the time delay for the location of leak point. Simulations show that VMD method can remove the noise under the premise of preserving the essential characteristics of the acoustic signal. The location precision can be obviously improved compared with the traditional EMD method.