基于欠定盲分离及局域均值分析的城市埋地燃气管道微小泄漏定位方法研究

项目名称： 基于欠定盲分离及局域均值分析的城市埋地燃气管道微小泄漏定位方法研究

项目编号： No.51204145

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 冶金与矿业学科

项目作者： 孙洁娣

作者单位： 燕山大学

项目金额： 25万元

中文摘要： 对城市高压及次高压埋地燃气管道的微小泄漏进行定位是确保管道安全运营的重要问题，本项目拟通过分析泄漏产生并沿管道本体传播的应力波信号实现管道微小泄漏定位，提出针对城市复杂情况的一种新的定位方法。研究内容包括：(1)基于盲源分离的应力波提取方法：研究低信噪比下独立分量分析去噪以及混合振动信号欠定盲分离方法，实现复杂噪声情况下应力波的有效提取；(2)应力波信号精确时延估计方法：研究基于局域均值分解后的包络、瞬时频率及能量分布特征的时延估计方法，提高时延估计精度；(3)分析应力波传播速度的影响因素，建立模态、频散、管道本体参数及其他主要因素对应力波传播速度影响的模型，采用理论分析与实验相结合的方法精确估计实际传播速度，综合建模以提高泄漏点定位精度。本项目研究成果将为管道微小泄漏的精确定位提供新思路，有助于提高泄漏信号的提取效率及泄漏点定位精度，具有重要的经济和社会效益以及广阔的应用前景。

中文关键词： 燃气管道泄漏检测；欠定盲源分离；稀疏分量分析；局域均值分解；泄漏点定位

英文摘要： To ensure the safe operation of the city high pressure and secondary high-pressure buried gas pipeline facilities, it is very important to locate the small leakage accurately. In this research I explore and analyze the stress wave which is generated by the leakage and propagates in the pipe body. Then the task of buried gas pipeline small leakage location is accomplished. It proposes a novel location method aimed at the urban complex circumstance. The research content includes: (1)stress wave extraction on the basis of blind source separation. Under low SNR circumstance , consider ICA (Independent Component Analysis) to remove the noise and then investigate the underdetermined blind separation method to separate the de-noised signals. It will achieve stress wave effective extraction under the complex noise situation. (2)stress wave precise time delay estimation. Study a accurate time delay estimation method based on analyzing the local mean decomposed signal's envelope , instantaneous frequency and energy distribution features. It will improve the estimation precision. (3) Investigate influence factors of the stress wave propagation velocity. Establish the model which express the interaction relationship between stress wave velocity with propogation mode, frequency dispersion, pipe parameters and the other main

英文关键词： natural gas pipeline leak detection；underdetermined blind source separation；sparse component analysis；local mean decomposition；leak localization