混合智能算法输电线路故障识别研究  

Research on Fault Identification of Transmission Line Based on Hybrid Intelligent Algorithm

作　　者：刘武斌 彭华厦[1] LIU Wu-bin,PENG Hua-xia

机构地区：[1]湖南工业大学,湖南株洲412008

出　　处：《新型工业化》2018年第5期12-18,27,共8页The Journal of New Industrialization

基　　金：湖南省教育厅科学研究项目(17C0471)

摘　　要：电力系统故障类型识别是对电力系统故障进行快速准确处理的重要保证。利用输电线路各类型短路故障时的电流和电压变化情况提取特征量,并对数据进行预处理,其中包括利用DBSCAN聚类算法排除噪声(如设备故障导致的错误数据)。完成预处理的数据将利用概率神经网络(Probabilistic Neural Network,PNN)进行短路故障的识别,而神经网络的分度密度SPREAD参数将获得遗传算法(genetic algorithm,GA)的优化,使分类准确率得到提高。最后通过算例验证了该方法的可行性、有效性。The identification of power system fault types is an important guarantee for quickly and accurately dealing with power system faults. The characteristic quantity is extracted by using the current and voltage changes in each type of shortcircuit fault of the transmission line. And the data is preprocessed, which includes the use of DBSCAN clustering algorithm to eliminate noise（such as erroneous data caused by equipment failure）. The pre-processed data will use the Probabilistic Neural Network（PNN） to identify short-circuit faults, and the neural network’s indexed density SPREAD parameter will be optimized by the genetic algorithm to improve the classification accuracy. Finally, an example is given to verify the feasibility and effectiveness of the method.

关 键 词：输电线路 短路故障分类 聚类分析 概率神经网络 遗传优化 

分 类 号：TM75[电气工程—电力系统及自动化] TP18[自动化与计算机技术—控制理论与控制工程]