大型复杂天然气管道系统实时动态模型泄漏检测技术研究

The leak detection of the natural gas pipeline is much more complex and difficult than the one for the liquid pipelines. The pipeline leak detection by dynamic modeling is the only suitable and effective CPM (Computational Pipeline Monitoring) method for the natural gas pipeline, based on the internal flow of the pipeline. But now it is suitalbe only for the simple single pipeline with a relatively slow transient flow. Accordingly, this study will start from the observation of the spread and development of leakage occurance along the pipeline, with consideration of the features of the large complex pipeline system and its measurement distribution. Consequently, the impact of the leakage occurance on the flowing distribution along the pipeline and the measurements on the pipeline ends will be revealed. Also will be considered seriously are the non-linear feature of the flowing model, the noises of the measurement and detection, the parameter uncertainty, and other random factors, which have great impact on the precise online simulation of the pipeline system. The detecting procedure involves the filter of the nonlinear system, parameter estimation and training, stochastic processes of the measurements, and so on. Here, the theories of the optimization and sensitivity analysis will be used to establish the pipeline realtime transient model (RTM) which has the capacities of the self adaptive (training) and the nonlinear filtering. The accurate online dynamic modeling will be achieved by the coupling of the study-observation procedures. Pursuant to the random state diagnosis and stochastic process analysis, and the correlation analysis of the simulated-measured pressure difference distribution, we will study and establish the new leak detection theories and methods on the basis of the precise online RTM model, which is essential and makes the pipeline leak detection insensitive to various noises and sensitive and quickly responsible to the occurrence of the tiny leakages. Also the achievement and estalishment of the study will realize the quick and reliable diagnosis and detection of the small leakages for the large complex natural gas pipeline systems. Consequently, the final issues of the study and research will setup the strong theoretic basis and foundation for the development of the applicable leak detection technology by the dynamic modeling for the natural gas pipelines. It will also provide the key and core technologies, practical models, methods and routines for the respective technology development and improvement, which is the independent intellectual property of China.

较之液体管道，天然气管道泄漏检测更加复杂和困难，动态模型泄漏检测是其唯一行之有效的基于管道内部流动的泄漏检测方法，但目前它仅适用于单根管道的缓变流动；本课题针对大型复杂天然气管道系统及量测数据分布特点，从泄漏传播和发展规律研究着手，研究管道流动参数分布和端点量测数据对泄漏的响应特征；考虑天然气管道流动模型的非线性及检测过程噪声、参数不确定性等随机因素影响，采用最优化和敏度分析理论，解决非线性滤波和参数估计问题，建立管道在线非线性滤波自适应动态仿真模型，通过学习-观测耦合，实现精确在线动态仿真；依据随机状态诊断和过程分析、动态仿真-实测压力分布差相关性分析等技术，建立动态模型泄漏检测新理论和方法体系，扩展动态模型泄漏检测的适应性,实现大型复杂天然气管道系统剧烈瞬变流动下微小泄漏的及时和可靠诊断，并在各种实验进行测试、验证和完善，为具有自主知识产权气体管道泄漏检测技术的开发和完善奠定理论基础。

本项目针对大型复杂天然气管道系统工艺与测量数据分布特点，首先研究了天然气管网系统在线仿真技术，实现了管网系统的实时仿真，获取了管道沿线实时压力和流量分布；其次，研究了基于最优化理论的滤波技术，减小了管道压力、温度以及流量信号量测噪声对仿真精度的影响；再次，研究了管道在线自适应仿真技术，实现了管道系统参数的自适应学习，提高了在线仿真精度；最后，研究了天然气管线动态模型压力分布泄漏检测技术，分析了不同位置、不同泄漏量对天然气管道系统压力分布的影响，以及泄漏规律。最终提出基于管道系统在线仿真-实测压力分布偏差相关性分析的泄漏检测技术，建立了动态模型泄漏检测新理论和方法体系，扩展动态模型泄漏检测的适应性，突破管道系统分段检测的限制，实现了大型复杂天然气管道系统剧烈瞬变流动下微小泄漏的及时和可靠诊断。. 在理论研究的基础上，项目组研发了管网仿真软件PNS V4.0和油气管道泄漏监测系统PIPELEAK，分别实现复杂管网系统的静态/动态仿真、离线/在线仿真和质量守恒法、压力节点分析法、参数分布法、动态模型法以及压力偏差相关性分析模型等泄漏检测理论的单独和综合运用。两件软件打破了国外公司在相关领域的垄断，在西安石油大学“管网仿真综合实验室”和中石化天然气分公司榆济管线上得到了充分验证和完善。该实验室以现代工业标准设计，由长输管道物理和虚拟现实系统组成，可实现油气管网模拟仿真、泄漏检测、系统分析、优化调控等功能，是油气管道系统高级教学和科研平台。. 研究成果已发表学术论文6篇，取得软件著作权3项，并在榆济天然气管道上成功应用。课题组同时与陕西燃气集团有限公司合作，成功申报陕西省科技统筹创新项目，将在靖西二线建立应用示范工程，形成推广应用中心和平台。此外，研究成果也在中石化川气东送管道、中石油北京调控中心等多家管道运行和管理单位进行测试，收到良好反馈。