【作者】 高华；

【导师】 王宝光；

【作者基本信息】 天津大学 ， 测试计量技术及仪器， 2009， 博士

【摘要】 天然气作为一种清洁、高效的能源,日益受到国内、外的重视。由于天然气的易燃易爆特性,作为天然气输送主要手段的管道的安全性显得尤为重要。随着我国能源结构调整进程的加快,城镇天然气管网的建设也在飞速发展。相对于长输管道,城镇天然气管网位于人口聚居地的地下,从城镇居民的安全角度看,其泄漏造成的危害性远高于长输管道。而城镇燃气管网的压力低,多分支,多附件、处于复杂多样的环境噪声中的特点,决定了其泄漏检测定位的难度远高于长输管道的泄漏检测定位。广义声发射是近些年发展起来的用于解释由泄漏引发的管壁中的应力波理论,该理论认为由泄漏引发的信号为连续信号,这种特性使得其在检测理论上有别于常规的负压波理论,也正是由于该信号的连续特性,可对其进行连续捕捉,而不必担心由于错过了时间而漏检。基于高阶累积量理论的高阶谱是信号处理领域中崭新的手段,具有基于二阶统计量的处理手段所无法比拟的优越性。文中将广义声发射理论和高阶谱理论结合起来,应用于城市地下燃气管网的泄漏检测及定位,主要工作有:1)按照国家标准,铺设了36m管线模拟城市地下燃气管网,其中直管线30m,过弯头管线6m;2)采用广义声发射理论解释由泄漏引发的管壁中的应力波现象,并据此理论研究了城市中(B)、低压燃气管网泄漏信号的特点,结合应力波在管壁中的传播特性,选择声频范围内的泄漏信号作为检测对象;3)在双谱理论的基础上,依据双谱可辨别信号平稳性的特点及统计独立的信号和的双谱等于信号双谱和的特性,研究了城市环境噪声——汽车、火车、施工等噪声的平稳性;4)依据双谱估计理论,研究了泄漏信号、管内流体噪声及检测仪器噪声的高斯性,依据双谱估计理论,采用了基于双谱的统计量,实现了在复杂城市环境噪声中的泄漏信号的检测及泄漏点的定位;5)在对两泄漏信号统计独立及检测噪声与信号统计独立的假设前提下,建立了两点泄漏模型,并利用四阶累积量对位于两检测系统之间的泄漏点数目进行了估计,实验结果表明,在信噪比较高的情况下,该方法对泄漏点数目的估计的准确率可达95%以上;6)在两点泄漏模型的基础上,从双谱理论出发,建立了两泄漏信号的自双谱和互双谱组成的方程组,巧妙的利用双谱域中的两条直线上的双谱值,建立了两泄漏点的定位双谱理论公式,并在实验的基础上,验证了该公式的正确性及该方法在工程上的可行性。该理论对检测范围内存在两泄漏点的情况的检测及定位具有很高的应用价值。大量现场实验结果表明:双谱技术可有效实现复杂城市环境噪声中的地下中压B及低压燃气管网的泄漏检测,在0.005～0.2Mpa的压力范围内,可检测最小距离为30m,定位绝对误差小于1.5m。更多还原

【Abstract】 As a kind of clean and efficient energy, Natural gas get more and more attention of the world. Flammable and explosive make the security of gas pipeline, which is the mainly transportation method, become much more important. With the adjustment of the energy source structure in China, urban underground gas pipeline is being developed. Campared to long-distance pipeline, the leak of urban gas pipeline is more dangerous. What’s more, due to the low pressure, multi-branch and layed in complex urban noise and so on, the detection and location of the pipeline leak are very difficult.Generalized Acoustic Emission is a developing theory which is used to explain the stress wave caused by leak of the pipeline. In this theory, the leak signal is considered as continuous signal. Unlike negative pressure wave theory, continuity of the leak signal will not fail to detect the leakage signal just for not capturing negative wave in time. Higher-order statistics, as a new signal processing method, has much more superiority to the general method based on two-order statistic. This thesis integrated generalized acoustic emission with higher-order statistics, and applied them to the leak detection of urban underground gas pipeline. The main works of this thesis are:1)According to the National Criterion, thirty-six-meter-long pipeline was layed. Thirty-meter-long pipeline was straight.2) Generalized acoustic emission theory was used to illuminate the stress wave phenomenon caused by the leakage. After studied the intermediate and low pressure of the urban gas pipeline, the audio frequency leakage signal was choosed as the detection target.3)Base on the third- and forth- cumulant and bispectrum, which can distinguish Gaussian and Nongaussian , stationery and non-stationery of the signal, the urban surrounding noise properties were studied. Bispectrum was applied to analyse the Stationarity of some peculiarly circumstance noise in urban, including vehicle noise, construction noise and so on.4)Leakage signal, flow noise and instrument noise were sorted in Gaussian and non-gaussian signal. Three statistics based on HOS (higher-order-statistics) were involved in the thesis to implement detection and location the leak.5)On the hypothesis of statistical independency among the two leakage signal and noise, two-leak-point modal was presented. Forth order cumulant was used to estimate source number between the two leakage detection systems. Experimental results show that the correctness ratio attained 95%.6) From the two-leak-point model, equation group composed by auto and cross bispectrum were used to location the two leakage event. Bispectrum on the two lines of the ( ) plane were involved. Experimental results show that method was available.Abuntant experimental results show that bispectrum applied to the urban middle or low pressure pipeline lied in complex surrounding is available. At 0.005～0.2Mpa, this method can detect and locate leak in 30m. The absolute location errors are less than 1.5m.更多还原