基于VMD和互谱分析的供水管道泄漏定位方法

Leak location method in water-supply pipeline based on combination of VMD and cross-spectrum analysis

针对低信噪比(SNR)下的供水管道泄漏振动信号用于时延估计泄漏定位误差大的问题,提出基于变分模态分解(VMD)和互谱分析结合的供水管道泄漏定位方法。首先,利用VMD将管道泄漏信号分解为若干个本征模态函数(IMF),对供水管道泄漏信号进行互谱分析确定特征频带;然后,利用IMF分量在特征频带内的能量比例作为选取准则来确定有效IMF分量,并对选取的有效IMF分量进行重构;最后,对重构信号进行时延估计来确定泄漏点位置。为了验证所提泄漏定位算法的有效性,通过仿真和实验分别对互谱与VMD结合、互相关及VMD与相关系数结合3种方法进行研究。实验结果表明,以上3种定位算法的平均相对定位误差分别为2. 53%,8. 62%和16. 86%。

Large leak location errors under the low signal-to-noise ratio(SNR)of leakage vibration signal in water-supply pipeline(WSP)are considered in this study.Leak location based on the combination of variational mode decomposition(VMD)and crossspectral analysis is proposed.First,the leak signal is decomposed into several intrinsic mode functions(IMFs)using VMD.The characteristic frequency band can be determined using cross-spectral analysis of the water supply pipeline leakage signal.Then,the effective IMF components are determined by using the energy ratio of the IMF component in the characteristic frequency band as a selection criterion.The selected effective IMF components are reconstructed to improve SNR.Finally,the reconstructed signals are used to estimate time delay to find a leakage position.To verify the effectiveness of the proposed method,the leak location based on combination of VMD and cross-spectral,cross-correlation,and combination of VMD and correlation co-efficient are compared in simulation and experiment respectively.Results show that the average relative location errors of the above three kinds of leak location algorithms are 2.53%,8.62%,and 16.86%respectively.