局域均值分解分析的管道泄漏孔径识别及定位

发布时间：2018-01-20 19:03

本文关键词： 管道泄漏孔径识别局域均值分解有效值熵支持向量机泄漏定位　出处：《仪器仪表学报》2014年12期 　论文类型：期刊论文

【摘要】：提出一种基于局域均值分解的RMS熵及时延估计的天然气管道泄漏孔径识别及定位方法。该方法首先对泄漏信号进行LMD分解,得到若干具有物理意义的PF分量并计算其有效值,进而结合信息熵的概念得出不同泄漏孔径的RMS熵,将不同孔径泄漏信号的多个RMS熵组成特征向量输入SVM进行识别。为提高互相关法定位精度,提出根据LMD分解结果的峭度特征进行重构再进行互相关获取时延信息,并结合泄漏信号的传播速度,实现泄漏点定位。实验结果表明该方法能够实现管道泄漏孔径有效识别及定位,且与基于EMD的RMS熵方法相比,识别效果更好,较直接相关法的定位精度明显提高。
[Abstract]:A method for identifying and locating leakage aperture of natural gas pipeline based on local mean decomposition (LMS) based on RMS entropy estimation in time is proposed. Firstly, the leakage signal is decomposed by LMD. Some PF components with physical significance are obtained and their effective values are calculated, and then the RMS entropy of different leak aperture is obtained by combining the concept of information entropy. In order to improve the accuracy of cross-correlation method, multiple RMS entropy vectors of different aperture leakage signals are input into SVM for identification. According to the kurtosis characteristic of the LMD decomposition result, the delay information is obtained by the reconstruction and cross-correlation, and the propagation speed of the leakage signal is combined. The experimental results show that the method can effectively identify and locate the leakage aperture of pipeline, and compared with the RMS entropy method based on EMD, the recognition effect is better. Compared with the direct correlation method, the positioning accuracy is obviously improved.
【作者单位】：燕山大学信息科学与工程学院;燕山大学河北省测试计量技术及仪器重点实验室;中国石油天然气管道通信电力工程总公司;
【基金】：国家自然科学基金(51204145) 河北省自然科学基金(E2013203300)资助项目
【分类号】：TE973
【正文快照】： 1引言随着天然气管网的迅速发展,管道由于老化、腐蚀、焊缝缺陷等原因而导致事故频繁发生,因此,开展对天然气管道泄漏识别与定位的研究,对保障管道安全运营具有重要意义。管道发生泄漏时,高速逸出的气体与泄漏孔壁发生气固耦合,产生应力波并沿管道本体向泄漏孔两侧传播,这种现

【参考文献】