基于LVQ神经网络的配电网故障定位方法

摘要

分析了配电网单相接地故障定位的难点及利用行波精确定位的问题.在利用故障点反射行波信号确定配电网单相接地故障距离的基础上,重点研究了应用学习向量量化(Learn Vector Quantization,LYQ)神经网络方法,利用该方法对不同分支的反射行波进行了特征提取与模式识别,实现了故障分支的判别,从而实现了精确定位.为了明确LVQ神经网络方法在精确定位方面的优越性,同时利用传统的BP( Back-Propagation)神经网络方法进行了比较.ATP-EMTP和Matlab软件对LVQ和BP两种神经网络方法进行了仿真验证；结果证明,在解决配电网单相接地故障精确定位问题方面,LVQ神经网络的效果优于BP神经网络.%According to the analysis on the difficulties of single phase grounding fault and the shortcomings of precise location using traveling wave signal in distribution network, a fault location method integrating the C-type of traveling wave location method with LVQ (Learn Vector Quantization) neural network is put forward. The aim is to combine different location methods to improve the accuracy of fault location. LVQ Neural Network is studied in this paper. The ability of feature extraction and pattern recognition is the characteristics of LVQ neural network, and it is applied to analyze reflected wave signals in different branches. In order to demonstrate the superiority of LVQ neural network, a classical BP (Back-Propagation) neural network has been developed to solve the same problem for comparison. The simulation results of ATP-EMTP and Matlab show that the LVQ neural network is quite effective and superior to BP neural network in fault location.