【作者】 苏广宁；

【导师】 严正；

【作者基本信息】 上海交通大学 ， 电力系统及其自动化， 2010， 硕士

【摘要】 在智能电网的大背景下,电网事故分析系统正在如火如荼的建设当中。本文的工作主要是事故分析系统的核心部分——故障诊断服务的研究与实现。基于IEC 61850的智能变电站有一套完整、系统的组织架构和支撑体系,这也是电网事故分析系统等高级应用的实现基础,本文首先对相关问题进行了概述。然后详细介绍了实际系统一次、二次设备的建模和导入内存库的过程,此为数据准备阶段。在此基础上利用内存库中的拓扑信息,结合前置采集的数据,通过系统结线分析划定故障区域。故障区域划定后,在主站端采用改进优化型故障诊断方法,进行诊断,并生成XML格式的诊断结果。本文对相关模型进行了改进,主要思想是保证诊断结果不发生漏判,并尽量减少误判。由于只基于保护、断路器状态量的故障诊断算法,在数据源方面存在先天的缺陷,所以引入录波数据进行信息融合,提高诊断精度,是一个必然的发展趋势。本文对信息融合理论做了比较详细的介绍,并对DS证据理论做出了一点改进。本文的信息融合考虑特征层和决策层两个层次。在特征层信息融合中,采用小波技术提取录波数据中的特征量,规约后得到点火序列,输入模糊Petri网,计算得到子站故障诊断结果,并对保护、断路器动作行为进行评价,形成故障简报上传主站,用以修正遥信量的错误。在决策层信息融合中,定义了三个专用于线路故障诊断的指标。采用各自不同的数据处理方法,将优化诊断结果和小波分析结果转化为对元件故障的支持度。考虑到多重故障的可能性,对以往在信息融合时将系统所有元件纳入一个辨识框架,令它们全体故障支持度之和为1的做法进行修改,将每个元件独立作为一个辨识框架。在本文提出的改进DS证据理论的方法下,将上述证据和优化结果进行融合,消除优化型方法可能产生的对线路的误诊断,得到精确诊断结果。整套方案在实时数字仿真系统(Real Time Digital System,RTDS)下得到了验证。更多还原

【Abstract】 In the background of Smart Grid, the Fault Analysis System is in full swing. The work of this paper is mainly about the core of this system -- the research and implementation of fault diagnosis service.The smart substation based on IEC 61850 has an integrated systematic organizational structure and support system, which is the basis of the realization of advanced applications such as Fault Analysis System. This paper overviews the related problems firstly.Then it is detailed introduced the process of primary and protection equipments modeling and imported into In-Memory Database, which is the stage of data preparation. Based on the topology information in the In-Memory Database and the data collected from the pre-collection equipment, the fault area is demarcated through grid topology analysis. After demarcating the fault area, the diagnosis method of optimization model is used at primary station, which generates the result format XML. This paper improves the model in order to reduce the false diagnosis rate under the premise of no missing one.Because of the fault diagnosis algorithm based on relay protection and breaker only has its inborn defects, it is a tendency to make information fusion with fault record data in order to enhance the effect of diagnosis. This paper introduces the information fusion theory in detail and makes a little improvement. The fusion is accomplished on both characteristic and decision layer.In the characteristic layer fusion, with the help of wavelet techniques, the firing sequence of Fuzzy Petri net can be extracted from fault record data. The substation fault diagnosis results as well as evaluation of the operations of protections and breakers are obtained through FPN calculation. Then a brief fault report is generated and sent to the primary station, which is used to revise the telesignal data.In the decision layer fusion, it defines three indicators, which is specially designed for line diagnosis. Different data processing methods are used to translate the optimization and wavelet analysis results into the support degree of components. Considering the possibility of multi-failure, it defines a frame of recognition for each element, revising the traditional method which adds up all of the components to one. Based on DS evidence theory improved by this paper, the aforementioned evidences are fused with optimization result. The turnkey solution can eliminate fault line diagnosis of optimization method, which is proved based on Real Time Digital System (RTDS).更多还原