管道泄漏检测中的退化多源信息融合模型研究

The pipeline leakage detection method based on the negative pressure wave is widely used, while the application effects are restricted because of its high false alarm rate. It is the basic way to remove the pipe leakage false alarm that making full use of the equipment state information from pipeline SCADA system to determine the source of the abnormal pressure. In this project, the water attack, which is also called transient flow, is used as the target and the information fusion system model of the degradation hydraulic system is constructed. The detection, tracking and attribute estimation of the target are then studied. The detailed research topics include: multi-field coupling effect rules in the oil pump station, degradation behavior model of the water attack in the pipeline, the multi-targets tracking method, the dynamic estimate on the target attributes and the information fusion models in the feature level and decision-making level. This project aims to study the building method of multisource information fusion model on degradation system, which provides the theoretical basis for solving the problems of the high false alarm rate and expanding the application field of information fusion.

基于负压波法的管道泄漏检测系统应用广泛，但其误报警率高的事实，制约了这一方法的应用效果。充分利用管道SCADA系统中的多种设备状态信息，来判断压力异常的产生根源，是排除管道泄漏检测误报警的根本途径。本项目以工况操作所引起的管内水击（瞬变流）为研究对象（目标），构建了水力系统的退化信息融合系统模型，对目标的检测、跟踪与属性估计方法展开了研究，具体内容包括：输油热泵站内多场耦合作用规律研究、长输管道内水击退化行为模型研究、多水击跟踪方法研究、水击属性动态估计方法研究、特征级和决策级信息融合模型构建等。本项目旨在研究退化特性系统的多源信息融合系统模型构建方法，为解决管道泄漏检测误报警率高的难题和拓展信息融合理论的应用领域奠定理论基础。

基于负压波法的管道泄漏检测系统应用广泛，但其误报警率高的事实，制约了这一方法的应用效果。统计发现生产现场管道泄漏监测系统的绝大多数误报警来自启、停泵操作所引起的水击波动。本项目以停泵所产生的正、负压水击和启泵所产生的正、负压水击为研究对象，按照数学模型假设、仿真数据获取、模型求解、模型验证这一范式，研究了泵站内水击持续时间属性的生成规律与泵站间管道内水击持续时间属性的传播退化规律；研究了管线上任意监测点处多个水击属性的信息融合模型。通过研究，本项目建立了长输原油管道中启、停泵操作下正、负压水击的持续时间属性与管长、管径、流体密度、粘度等参数的数学关系模型；提出了多水击持续时间属性的信息融合规则与其遵循的计算定律；在此基础上构建了新的管道泄漏监测系统模型。本项目的研究成果增加了对水击时间维度的认知，丰富了多水击信息融合模型理论，并为降低工程现场管道泄漏监测系统的误报警率提供了模型依据。