【作者】 杨海晶；

【导师】 杨以涵； 刘文颖；

【作者基本信息】 华北电力大学（河北） ， 电力系统及其自动化， 2009， 博士

【摘要】 电网故障诊断课题的定位是依据故障综合信息,借助于知识库或模型,采用某种诊断机制来确定电网故障元件及分析故障原因,同时完成对保护装置、安全自动装置等监测、控制设备的动作行为的评价。目前的故障诊断方法在全局化和实用化中仍存在一些问题,主要原因主要有以下几点,一是信息不全,二是信息噪声干扰大,三是诊断方法在信息不完备的情况下鲁棒性较差。为此,论文对基于柔性SCADA系统的电力系统复杂故障诊断方法进行研究。提出了信息交互性的柔性SCADA系统;在报警信息的处理方面,开发了智能报警系统;在此基础上,本文结合Petri网和专家系统的特点,提出了基于正反向推理的Petri网故障诊断模型,并用WAMS信息对故障信息和诊断结果进行验证。本文首先对信息完备化做了深入的研究,提出建立基于广域网的柔性SCADA系统,对故障信息按重要程度进行分级上传,同时具有主动上传和等待召唤两种方式,为故障诊断提供灵活、有效的数据交换和功能协调的平台。柔性SCADA系统不再是点对点的单向固定路径的信息交互,而是能实现网络中任意两个(或多个)节点间双向的信息交互,体现子站和主站信息交换的智能性,它是对传统SCADA的扩充和发展,拓宽了系统的数据来源,为故障诊断模块丰富了数据基础。其次,针对电力系统发生大面积复杂故障后,报警信息存在大量的噪声干扰问题,对报警信息处理和智能报警进行了研究,采用数据挖掘的方法,从历史数据中挖掘出规则,去除实时报警信息的噪声,在此基础上,提出用正向产生式推理,优先和分级报警,减少报警信息数量并提高报警质量和正确率。同时,将去噪后的信息提供给故障诊断模块进行故障分析,提高故障诊断信息的正确率。最后,本文提出了基于柔性SCADA系统的电力系统复杂故障诊断方法,对全局分层信息的电网故障诊断模型进行了研究。充分利用柔性SCADA的交互性特点,实现了报警信息的分层传输和分层利用,满足了简单故障的快速诊断和复杂故障的准确诊断要求。对于复杂故障,设计了应用元件Petri网故障诊断模型正向推理来确定故障元件方法。当报警信息不完全正确时,提出了应用元件保护配置时延Petri网进行报警信息纠错处理方法,提高容错能力。本文还提出了引用WAMS数据对报警信息纠错处理结果和故障诊断结果进行验证的方法,可进一步提高故障诊断的可靠性和正确性。经算例仿真验证了本文方法的有效性。更多还原

【Abstract】 The issue of fault diagnosis of power system is a kind of research that based on fault comprehensive information and Knowledge-Based System or model, using kind of diagnosis mechanism to determine fault equipments or reason, at the same time appraising the actions of protecting devices, automatic security devices and other such as automatic monitoring and controlling equipments. At present, the fault diagnosis methods still have some problems in overall situation and in practical. There are three main reasons. First, the information is incomplete. Second, the information noise interference is too large. Third, the robustness of diagnostic is too poor while the information is incomplete. To this end, the paper research about complex fault diagnosis method based on the flexible SCADA in power system. A flexible SCADA system that information can exchange is put forward. In the alarm information processing, intelligent alarm system is developed. After that, Petri nets fault diagnosis model based on forward and backward reasoning is put forward, combining the features Petri nets and expert system. The fault information and diagnostic results is verified by using WAMS information.First of all, the paper did a deep research to complete the information and raised to create the flexible SCADA system based on wide area network, which can classify and upload according to importance of fault information, having initiative and waiting two patterns at the same time, in order to provide a flexible and effective platform of data exchange and functions coordination for fault diagnosis. The flexible SCADA system is no longer a fixed point-to-point one-way path of information exchange, but the network can achieve any two (or more) nodes of information exchange in two-way, which can reflect intelligence of information exchange between master- station and the sub-stations. It is the development and expansion of traditional SCADA, expanding the system’s data sources for the fault diagnosis model.Secondly, as there are a lot of noise problems of warning messages after large-area complex fault in power system, information processing and intelligent alarm was studied in this paper. Noises of real-time alarm information are removed from historical database using data mining methods. On this basis, forward and backward reasoning is put forward to have priority classification alarm and reduce the number of alarm information and improve the quality and accuracy of alarm. At the same time, the information after the de-noising will be provided to the fault diagnosis system, so that the accuracy of fault diagnosis is improved.Finally, complex fault diagnosis method based on flexible SCADA is brought forward, as to study the fault diagnosis model for the overall hierarchical information. This method takes full advantage of flexible SCADA interactive features to achieve the stratification of transmission and use for alarm information, which meets the requirement of rapid for simple faults and preciseness for complex faults. As for complex faults, the Petri net forward reasoning model is designed to determine the fault components. When the alarm information is not completely correct, application components protecting configuration of delaying Petri nets is used to correct the error alarm information to improve fault tolerance. Besides that, this paper also made a way to test and verify the correcting alarm information and fault diagnosis using WAMS information. The simulation example proved the effectiveness of the method.更多还原