摘要
为了准确地检测城市燃气管道泄漏,提出了一种基于广义概念的管道泄漏检测定位方法。声发射技术对于管道泄漏的检测、定位是一个极好的工具,但由于泄漏源的传播容易受到周围背景噪声以及复杂工况的影响,其定位误差较大。基于时延估计的互相关信号处理方法被广泛用于管道泄漏检测定位,但由于泄漏应力波传播通道的动态特性,使得源信号在传播过程中会产生波形变化,给互相关函数峰值位置的确定带来困难。由此引入广义相关分析方法,通过对信号进行前置滤波,在一定程度上减少了传播通道动态特性因素对泄漏点定位的不利影响,得到了更为准确的时延估值。在此基础上,通过模拟实验,编写Matlab神经网络代码,构造GRNN模型,进一步预测定位。结果表明,GRNN预测的声发射检测值、互相关定位值以及广义相关定位值,相比之前定位精度分别得到提高,其中基于广义相关的延时估计方法定位最为精确,将该方法用于工程实际中,可以更加精确地定位出泄漏点。
In order to accurately detect leakage of city gas pipeline, a method is proposed based on broad concept of leak detection and location of pipeline. Acoustic emission technique is an excellent tool for the pipeline leak detection and loca- tion, but because the spread of the leak source is susceptible to the surrounding background noise and the impact of complex conditions, its positioning error is greater. The methods based on delay estimation cross - correlation signal processing are widely used in pipeline leak detection and location, but due to the dynamic characteristics of leakage stress wave propagation path, the source signal in the communication process will produce changes in wavefoml and the peak position to cross - cor- relation function determination difficult. Thus the generalized correlation analysis is introduced. To a certain extent, the ad- verse effects of the dynamic characteristics of the propagation path of the leak location factors are reduced by pre- filtering the signal and a more accurate estimate of the delay is achieved . On this basis, through simulation experiments, the matlab neural network code is compiled and GRNN model is constructed to further predict location. The results show that, GRNN predicted value of acoustic emission detection, cross - correlation values and generalized targeting positioning value are im- proved, compared to the previous positioning accuracy, among which the delay estimation based on generalized method of lo- cating is most accurate, if the method is used in engineering in practice, it will more accurately locate the leak.
出处
《工业安全与环保》
北大核心
2016年第1期80-84,共5页
Industrial Safety and Environmental Protection
基金
江苏省科技项目(BE2014625)
常州市科技项目(CE20145054)
2013年国家安监总局安全生产重大事故防治关键技术科技项目(安监总厅科技(2013)140号)
关键词
管道泄漏
互相关分析
广义相关分析
广义回归神经网络
pipeline leakage
correlation analysis
generalized correlation analysis
generalized regression neural network
作者简介
郝永梅，女，1970年生，副教授，硕士学位，主要研究方向：消防工程及油气储运风险分析。
徐明，硕士研究生，研究方向：油气储运安全。
