基于BP神经网络的配电网故障辨识方法

摘要

The feeder fault location method with fast and high tolerance characteristics plays an important role in enhancing the automation level of distribution network.With the wide application of terminal equipment for power distribution automation system, indirect positioning method based on information equipment over current fault section is simple and convenient and has become a hot spot of research in this field, the algorithm is mainly divided into two categories: unified matrix algorithm and swarm intelligence algorithm.Unified matrix algorithm for fault tolerant ability is poor, swarm intelligence methods in logical modeling of bottleneck problems existed during the optimization goal.Based on BP neural network model of distribution network fault location method using the characteristics of the FTU realize fault location, the principle is simple, easy implementation, and has multiple fault location.The simulation results prove the effectiveness of the proposed method.%随着配电网自动化终端设备的广泛应用,基于设备过电流信息的故障定位方法因原理简单、实现便捷而成为该领域的研究热点.该方法主要分为两类:统一矩阵算法和群体智能算法.统一矩阵算法的容错能力差,而群体智能算法在构造优化目标时存在逻辑建模的瓶颈问题.为有效克服上述缺点,提出了基于BP神经网络模式的配电网故障定位方法,利用FTU的特征量和BP神经网络的自学习性及良好的泛化能力实现配电网的故障定位,不但原理简单、实现方便,而且具有多重故障定位能力,仿真结果证明了该方法的有效性.