【作者】 孙洁娣；

【导师】 靳世久；

【作者基本信息】 天津大学 ， 精密仪器及机械， 2009， 博士

【摘要】 随着我国国民经济持续高速发展,对能源特别是石油资源的需求越来越迫切。管道运输作为运送石油的重要手段在国民经济中的战略地位愈显重要。由于管道输送介质的危险性和污染性,一旦发生事故将造成巨大的生命财产损失及环境污染。近年来我国的输油管网严重受到的破坏95%以上是由于人为侵入造成的,常规的泄漏检测技术及已有的检测装置主要通过管道输送压力、流量以及温度等参数的变化来判断泄漏是否发生,监测灵敏度不高且均是事后报警,能够在管道安全受到威胁或者正被破坏而尚未造成损失时就及时发现入侵行为是急待解决的问题,因此研究能够对管道安全进行监测预警的技术非常必要。本文研究了基于地震动信号的管道安全监测预警中的一些关键技术。通过实验采集目标在管道周围活动时产生的地震动信号,对信号进行预处理、特征提取和模式识别,从而确定引起震动的目标性质,并判断目标是否对管道构成入侵威胁,对于非法入侵目标能够进行定位。本文主要进行了以下几方面的研究工作:1、首先研究了利用地震波进行地面目标活动监测的机理,分析了地层中多种类型的地震波以及它们的传播机理,确定了利用瑞利波进行地面目标活动探测的方法,并设计了实验装置采集了多种目标的信号。2、地震动信号是典型非平稳信号,本文在研究非平稳信号特性及分析方法的基础上,提出了采用基于小波包能量谱的信号特征提取方法、基于经验模态分解的信号特征提取方法以及基于高阶统计量的信号特征提取方法用于地震动目标信号分析中,提取目标特征。3、分析了管道安全监测预警中采用多传感器进行目标信号采集及监测的方法。在研究多传感器数据融合方法的基础上,采用了基于证据理论的融合方法进行目标识别结果的融合处理,识别准确率较单一传感器结果有明显提高。4、研究了目标定位的原理以及基于信号到达时间差(TDOA)的定位方法。在TDOA方法的基础上,针对地震动信号非平稳的特征提出了一种新的定位方法:基于希尔伯特黄变换的特征频率及到达时间差的目标定位方法,实验表明方法是有效的。更多还原

【Abstract】 The sustained high-speed national economic development in China makes the demand of energy supplies especially oil resources increase rapidly. The oil Pipeline transportation plays an extremely important role in the national economy. Because of the dangers and pollution of the pipeline transported medium, the leakage accidents may cause huge loss of life and properties along with the environmental pollution. In recent years, the national oil transportation pipeline networks were seriously damaged; and ninety five percent of them were caused by the artificial invasions. Presently the prevalent methods to judge leakage are evaluation of parameters, such as pressure in pipeline, flow rate and temperate obtained by common oil and gas pipeline leakage detecting devices to infer whether a leakage has occurred. The monitoring sensitivities of these methods are low and the warning of leakage exclusively arrives after the leakage. The serious problem which urgently needs to be solved is to detect the pipeline invasion actions and avoid them before the pipeline security was threatened or the damages were occurring caused by the invasions but the loss doesn’t cause. Therefore it is necessary to study the pipeline security monitoring and pre-warning technology.This dissertation involves the in-depth study on the pipeline security monitoring and pre-warning technique based on seismic signals. The seismic signals generated by the different targets around the pipeline were gathered. These signals were pre-processed and the targets features were extracted. The pattern recognition was done to identify the targets. The next work is to judge whether the targets belongs to the invading class and the localization is finished.The major study of this dissertation covers the following aspects:1. It studies the monitoring principle which utilizes the seismic signals to detecting the motion targets and also elaborates a variety of seismic waves in the stratum and the dissemination mechanism. It confirmed the method which makes use of the Rayleigh wave to detecting the ground moving targets. The experiments were designed to gather the several different target signals.2. The seismic signals are typically non-stationary. It analyzed the characteristics and analyzing methods of non-stationary signals. It applies the energy spectrum signal feature extraction method based on the wavelet package decomposition, the EMD (Empirical Mode Decomposition) based signal feature extraction method, and the signal feature extraction method based on the higher-order statistics to extracting the eigenvectors of the detected signals along the pipelines.3. It studies the target signals acquiring and detecting methods based on multi-sensors in the pipeline security monitoring and pre-warning technique. On the basis of researching the multi-sensor data fusion methods, it proposed the evidence reasoning theory to dispose the single recognition results. The fashioned outcomes are better than these single recognition results.4. It investigated the targets localization principles and the method based on the time difference of arrival (TDOA). On the basis of TDOA and In view of the seismic signals’non-stationary characteristics, it proposed a new method: a new localization method based on HHT (Hilbert-Huang Transform) characteristic frequency and time difference of arrival. The experiment verifies the new method is effective.更多还原