【作者】 王喜良；

【导师】 付光杰；

【作者基本信息】 大庆石油学院 ， 电力电子与电力传动， 2004， 硕士

【摘要】 原油管道运输中的泄漏事故，不仅损失原油和污染环境，还有可能带来重大的人身伤亡。近些年来，管道泄漏事故频繁发生，为保障管道安全运行和将泄漏事故造成的危害减少到最小，需要研究泄漏检测技术以获得更高的泄漏检测灵敏度和更准确的泄漏点定位精度。管道泄漏检测是多领域、跨学科的课题，涉及到管道流体力学、热力学、传感技术、微弱信号检测、信号处理等多个学科，本文对泄漏检测及定位技术进行了初步的研究，主要进行了以下几方面的工作： 1．分析了负压波泄漏定位方法的两项关键技术，提出了消除首末、端压力传感器动态响应时间差的方法，采用GPS来统一首末端数据采集系统的系统时间；使定位更加准确。利用信号奇异点与小波变换模极大值在多尺度上变化对应的性质，将离散小波变换应用到获取压力波信号序列特征奇异点中，可以对泄漏点进行定位。因为现场采集的数据中含有大量噪声，我们采取小波门限去除噪声干扰的方法，可以很好的消除噪声干扰。在实际应用中取得了良好的效果。 2．本文将基于小波包分析的“能量—故障识别”故障诊断方法引入管道运行状态监测技术，提出了反映泄漏压力信号特性的特征向量指标。该方法对信号进行小波包分解，通过分析在每个分解节点上重构的新序列，就能根据各频带内的特征向量对压力变化原因进行识别。 3．本文应用正常情况下的特征向量与泄漏情况的特征向量进行比较，并且采用阀值的方法判断泄漏故障的发生。本文还采用RBF神经网络的方法对对管道的运行状态进行在线识别，可以检测出管道的泄漏，仿真结果表明效果较好。 4．最后应用网络控制理论，建立了一套输油管线泄漏检测定位的网络控制系统，并给出了相应的网络控制模型和基本的软硬件实现步骤。更多还原

【Abstract】 The leakage accidents in pipelines transportation not only cause great loss of material, but also pollute the environment; what’s more, it may lead to serious casualty. Leakage accidents in pipeline transportation occur frequently in recent years. In order to guarantee the pipelines work safely and minimize the losses caused by leakage accidents, it is necessary to study leakage detecting technology to raise the sensitivity of detection and accuracy of localization. Pipeline leakage detection is a multi-field object, which involved many subjects, such as pipeline hydrodynamics, thermodynamics, sensor technology, subtle signal detection and signal processing. This search on pipeline leakage detection and localization is carried out in this dissertation The main efforts accomplished can be summed up as the following aspects:1. Two crucial technologies in instantaneous negative pressure wave method are analyzed in this dissertation, then a means is presented which can eliminate the dynamic response time diversity between pressure sensors in the beginning and end of the pipeline, by adopting GPS to unify the system time between data collecting systems in two ends. Taking account of the corresponding characteristic of the signal singularity and wavelet transform module maximum in multi-scale changing, the discrete wavelet transform is applied to get the featured inflection point of negative pressure wave. Because data that scene gather contain a large number of noises, we use wavelet valve method denoise interferes, Have made the good result in practical application 2. "Energy -fault recognition", a fault diagnosis method based on wavelet packets analysis for the transport pipeline supervising system is proposed in this dissertation. Eigenvector target is presented which reflect the characteristic of leakage pressure signal. The signal is wavelet packets decomposed in this method. By analyzing the new sequences reconstructed on every decomposition note, the fault then can be recognized according to the eigenvectors in each frequency band.3. Use characteristic vector quantity of normal situation with leak characteristic vector quantity of situation compare, adopting valve method determinant the leakage of pipelines. The method that this text still adopts RBF neural network discerns the state of pipeline online, Can measure out the leakage of pipeline , the simulation result indicates that the result is better.4. Use network control theory, set up one petroleum pipeline leak network control system of leakage detection and localization, corresponding network control models and basic software and hardware had been put out.更多还原