基于免疫RBF网络改进小波分析的小电流接地故障选线  

Fault line detection in indirectly grounding power system based on wavelet analysis improved by immune RBF network

作　　者：陈少华[1] 尹胜兰[1] 莫哲[1] 

机构地区：[1]广东工业大学自动化学院,广东广州510006

出　　处：《电力系统保护与控制》2012年第3期46-50,共5页Power System Protection and Control

基　　金：广东省教育厅电力节能与能源技术重点实验室资助项目(IDSYS200701)

摘　　要：配电网发生单相接地故障,当故障时刻处在故障相的相电压过零点且大电阻接地故障时,基于小波分析选线会因故障线路与非故障线路暂态量差别不明显而发生误判。提出基于免疫RBF网络改进小波分析算法,通过免疫机制对不同故障类型故障线路与非故障线路的暂态零序电流小波模极大值进行聚类,确定RBF网络隐层中心。通过遗传算法训练RBF网络得到最优连接权值。将训练后的RBF网络用于的小电流接地故障选线,仿真结果表明,利用该算法选线具有较高的精确度。When single-phase-to-earth fault occurs at faulty phase voltage crossing zero and with big ground resistance in distribution network, the faulty line detection will be misjudged due to the unobvious difference of transient information between fault line and non-fault line based on line detection of wavelet analysis. A fault line detection based on wavelet analysis improved by immune RBF network is proposed. First, wavelet modulus maxima of fault line and non-fault line transient zero-sequence current is clustered at different fault types by immune mechanism. It determines hidden layer centers of RBF network. Then the RBF network is trained with genetic algorithm to obtain optimal weight value. The trained RBF network is applied to fault line detection in indirectly grounding power system. Simulation result indicates that the faulty line detection by the presented method is more precise.

关 键 词：免疫 RBF网络 小波分析 零序电流 小电流接地 

分 类 号：TM862[电气工程—高电压与绝缘技术] TP183[自动化与计算机技术—控制理论与控制工程]