摘要
针对10kV配电网单相接地短路故障发生位置难以确定的问题,提出利用配电网中的多源数据,通过构建分层结构神经网络进行故障区段定位的方法。首先提取各种类型10kV配电网的固有特征,在分层结构神经网络的聚类层根据配电网的这些固有特征,利用自组织映射神经网络进行聚类分析,得到不同类别配电网;然后在分层结构神经网络的训练层对各类配电网分别用广义回归神经网络对故障定位条件特征与结果特征进行训练,得到各类配电网的故障定位模型;最后将发生单相接地短路故障的配电网下属各区段故障定位条件特征输入至所对应的故障定位模型中,判断各区段故障情况,实现故障定位。实际算例分析表明,所提出的方法能快速、准确地找出10kV配电网单相接地故障发生的区段,且具有较好的容错性。
To solve the problem of 10 kV power distribution network short fault positioning,a method based on multiple data source and multi-layer neural network was proposed.Firstly,inherent features of different 10 kV distribution network was extracted.Self-organizing map in multi-layer neural network was used to divide them into different categories in the clustering layer.For each category,general regression neural network was adopted to train the fault positioning features and get positioning model.Finally,the features of the power units in which the short fault happens was input into the positioning model to get the result of the position where the phase to earth fault happens.The practical analysis shows that the method can find the position of 10 kV power distribution network short fault with quickness and accuracy.Meanwhile,it has great fault tolerance.
作者
苏运
刘思怡
张焰
SU Yun;LIU Si-yi;ZHANG Yan(State Grid Shanghai Municipal Electric Power Company,Shanghai 200437,China;Department of Electrical Engineering,School of Electronic Information and Electrical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)
出处
《水电能源科学》
北大核心
2019年第11期176-179,共4页
Water Resources and Power
基金
国家电网公司科技项目(520940170020)
