基于核Fisher判别和神经网络的输电线路故障诊断研究

发布时间：2018-03-18 03:01

本文选题：故障诊断　切入点：核Fisher特征矢量　出处：《安徽理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：输电线路是整个电力系统的传输枢纽,承载着输送电能的作用,要求必须具有高可靠性。然而,整个电力系统最容易发生故障的环节也是输电线路,其发生故障的原因多种多样,随机性强,而且包含很多非线性故障问题,很难及时诊断出。如何保证输电线路安全、高效、准确地运行,并且在故障发生时快速诊断,成为研究人员关注的重要问题。目前,应用单一的人工神经网络算法,很难处理好这一非线性问题。所以,本文介绍了一种新的研究方法,即将核Fisher判别法与人工神经网络相融合的方法,既能最大程度的利用人工神经网络的优点,又能够高效的处理一些非线性故障信息。这一研究方案给工业生产和科研探索带来极大的便利,具有很大的实用价值和现实意义。本文首先介绍分析了输电线路上经常发生的一些故障情况以及对应的故障机理,然后选取合适的核函数,输入样本信息和需要处理的数据,借助非线性函数关系将其从自身的空间投射到高维空间里面。经过进一步数据处理,获得最优的核Fisher特性向量解。最后,通过检测参数确定表征故障的特征量,进一步确定输电线路中的故障类型。针对目前短路故障中信息处理时的复杂性,以及单一诊断方法的实时性和准确性差等问题,构建了基于KFDA的输电线路故障诊断模型,最后在MATLAB上进行基于核Fisher的输电线路故障检测仿真分析。本文采集大量的样本数据,进行对比分析后,证明了所设计的输电线路诊断模型的可行性及准确度。结果表明,本文所确定的基于核Fisher判别和神经网络故障诊断方法能很好的判别输电线路故障短路类型,与预期的目标基本相同。
[Abstract]:The transmission line is the transmission hub of the whole power system, which carries the function of transmitting electric energy and must have high reliability. However, the transmission line is also the most prone to failure of the whole power system. The causes of fault are various, randomness is strong, and many nonlinear fault problems are included. It is very difficult to diagnose in time. How to ensure the transmission line to operate safely, efficiently and accurately, and to diagnose quickly when the fault occurs. At present, it is very difficult to deal with this nonlinear problem by using a single artificial neural network algorithm. Therefore, a new research method is introduced in this paper. The combination of kernel Fisher discriminant method and artificial neural network can make maximum use of the advantages of artificial neural network. It can also deal with some nonlinear fault information efficiently. This research program brings great convenience to industrial production and scientific research. It is of great practical value and practical significance. This paper first introduces and analyzes some faults often occurring on transmission lines and the corresponding fault mechanism, then selects the appropriate kernel function, inputs the sample information and the data to be processed. With the help of nonlinear function relation, it is projected from its own space to high dimensional space. After further data processing, the optimal solution of kernel Fisher characteristic vector is obtained. Finally, the characteristic quantity of fault is determined by detecting parameters. In view of the complexity of information processing in the current short-circuit fault and the poor real-time and accuracy of the single diagnosis method, the fault diagnosis model of transmission line based on KFDA is constructed. Finally, the simulation analysis of transmission line fault detection based on core Fisher is carried out on MATLAB. After collecting a large number of sample data, the feasibility and accuracy of the designed transmission line diagnosis model are proved. The kernel Fisher discriminant and neural network fault diagnosis method based on this paper can distinguish the fault short circuit type of transmission line very well, which is basically the same as the expected target.
【学位授予单位】：安徽理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM755

【相似文献】