基于广义S变换的长输管道泄漏检测和定位方法  

Long Pipeline Leak Detection and Location Method Based on Generalized S Transform

作　　者：雷阳 

机构地区：[1]福建省计量科学研究院,福建福州350003 [2]国家蒸汽流量计产品质检中心,福建福州350003

出　　处：《质量技术监督研究》2016年第3期2-8,共7页Quality and Technical Supervision Research

基　　金：国家自然科学基金(No.60974039)

摘　　要：基于广义S变换,研究了长输管道泄漏检测与定位方法。首先,针对泄漏信号的识别易受工况调节信号干扰的问题,对各类信号进行广义S变换时频分析,提取相应的特征向量。采用概率神经网络(PNN)作为分类器,对泄漏信号进行识别。然后,提出了一种基于相关性分析的广义S变换方法,精确获得泄露负压波到达管道首末站的时间差。利用广义回归神经网络(GRNN)拟合泄漏定位公式,确定泄漏点的位置。最后,通过对实际输油管线现场数据进行分析,验证了文中所提出方法的有效性。The long pipeline leakage detection and location problem was researched based on the generalized S transform in this paper.Firstly,for the problem that the identifi cation of leakage signals is apt to be affected by regulatory signals,the general S transform was used to extract the feature vectors of the signals.And the leakage signals were identifi ed by the probabilistic neural network（PNN） classifi er.Then,a generalized S transform method was proposed based on the correlation analysis.The time difference of the leakage negative pressure waves between upstream and downstream was acquired accurately by using this method.The formula of leakage location was fi tted by the generalized regression neural network（GRNN） and the leak position was located.Finally,the effectiveness of the proposed method was verifi ed by analyzing the on-site data of the real oil pipeline.

关 键 词：广义S变换 概率神经网络 广义回归神经网络 泄漏检测 泄漏定位 

分 类 号：U178[交通运输工程]