基于支持向量机的电网故障诊断研究

发布时间：2018-03-26 00:29

本文选题：电网故障诊断　切入点：复杂故障　出处：《华北电力大学(北京)》2017年硕士论文

【摘要】：随着全球能源互联网的提出,各级电网之间的互联性增强,尤其是间歇式新能源和微电网的接入,电网自身的动态行为日趋复杂,发生拒动等复杂故障的可能性仍旧存在,准确、快速的实现电网故障诊断成为更为紧迫的现实要求。本文以诊断电网复杂故障为目的,针对现有专家系统缺乏学习能力的不足,利用支持向量机(SVM)模型基于历史故障经验对复杂故障进行诊断,其主要工作如下:研究统计学习理论的内在特点,深入理解VC维和结构风险最小化原则,阐述支持向量机模型的数学原理,在Visual studio 2010平台使用C++语言分别编写线性和非线性支持向量机模型的实现算法,为支持向量机在电网故障诊断的应用打下理论基础。根据电网故障诊断的需求与特点,提出基于支持向量机的方法对复杂故障进行诊断。通过对历史故障案例的训练学习,获取这些复杂故障案例中的“隐性”诊断知识,利用这些经验规律对现有故障进行诊断,并不断用新的故障事件对SVM模型进行再训练。为使建立的支持向量机模型具有广泛的通用性,以数据采集与监视控制系统(SCADA)系统采集的保护动作和断路器跳闸信息为基础,针对母线、线路和变压器等三类元件分别设置其支持向量机模型的输入特征量,并利用遗传算法对惩罚因子和径向基核函数参数进行寻优。在课题组已开发的电网故障诊断系统的基础上,增设利用SVM算法的复杂故障诊断模块,并对该模块五个组成部分的功能设计进行了逐一论述,提高了电网故障诊断系统针对复杂故障的诊断效率,最后通过三种典型的复杂故障案例测试验证了模块实用性与有效性。
[Abstract]:With the development of the global energy Internet, the interconnection between all levels of power grids is enhanced, especially the interconnect of intermittent new energy sources and microgrids, the dynamic behavior of the grid itself is becoming more and more complex, and the possibility of complex faults such as rejection still exists. Accurate and rapid realization of power network fault diagnosis becomes a more urgent practical requirement. This paper aims to diagnose complex fault of power grid, aiming at the lack of learning ability of the existing expert system. Support vector machine (SVM) model is used to diagnose complex faults based on historical fault experience. The main work is as follows: the inherent characteristics of statistical learning theory are studied, and the principle of VC and structural risk minimization is deeply understood. The mathematical principle of support vector machine (SVM) model is expounded. The algorithms of linear and nonlinear SVM models are programmed in C language on Visual studio 2010 platform. This paper lays a theoretical foundation for the application of support vector machine in power network fault diagnosis. According to the requirements and characteristics of power network fault diagnosis, a method based on support vector machine is proposed to diagnose complex faults. Acquiring the "implicit" diagnosis knowledge in these complex fault cases, and using these empirical rules to diagnose existing faults, In order to make the established support vector machine model universal, it is based on the protection action and breaker tripping information collected by the data acquisition and monitoring control system (SCADAA) system. For three kinds of components, such as busbar, line and transformer, the input characteristic quantity of support vector machine model is set up separately. The genetic algorithm is used to optimize the parameters of the penalty factor and the radial basis function. On the basis of the fault diagnosis system developed by the research group, the complex fault diagnosis module using SVM algorithm is added. The function design of the five components of the module is discussed one by one to improve the efficiency of the fault diagnosis system for complex faults. Finally, the practicability and effectiveness of the module are verified by three typical complex fault cases.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM711

【参考文献】