配电网单相接地故障辨识方法

Identification of Single-phase Ground Fault

当发生单相接地故障时,一般允许故障下运行1～2h,此种情况下,一旦发生单相接地故障,如果不能及时切除,则故障极有可能进一步发展,对系统的安全造成更严重的威胁,所以当系统故障时,必须及时查明故障原因并消除故障。对中性点零序电压进行了小波分析,提取了各类故障的小波频带能量故障特征量。对BP神经网络的特性进行了分析,就存在的缺陷进行了优化,重点研究了利用PSO算法对人工神经网络进行优化,实现了PSO优化神经网络算法;通过对PSO优化神经网络进行训练和测试,将辨识结果与基于人工神经网络的故障辨识结果进行比较,验证了基于PSO优化神经网络算法的单相接地故障辨识更加准确和高效。

This paper is about the zero sequence voltage of neutral point is analyzed,respectively the wavelet frequency band energy of fault features are extracted.Then the BP neural network is studied.Characteristics of BP neural network are analyzed,and the emphasis is on the use of PSO optimization algorithm to optimize artificial neural network,then train and test based on PSO optimized neural network,compare the identification results of artificial neural network and PSO optimized neural network.