基于随机森林算法的配电线路短路故障分类  

Classification of Short Circuit Faults in Distribution Lines Based on Random Forest Algorithm

作　　者：刘树鑫[1] 卓裕 李津 田二胜 刘洋[2] 冯毅[3] LIU Shuxin;ZHUO Yu;LI Jin;TIAN Ersheng;LIU Yang;FENG Yi(School of Electrical Engineering,Shenyang University of Technology,Shenyang 110870,China;State Grid XJ Group Corporation,Xuchang 461000,China;State Grid Xingtai Power Supply Company,Xingtai 054001,China)

机构地区：[1]沈阳工业大学电气工程学院,辽宁沈阳110870 [2]国家电网许继集团有限公司,河南许昌461000 [3]国网邢台供电公司,河北邢台054001

出　　处：《电器与能效管理技术》2020年第8期52-57,67,共7页Electrical & Energy Management Technology

基　　金：国家自然科学基金(51977132);国网河北省电力有限公司科技项目(5204XQ190003)。

摘　　要：为实现对配电线路短路故障后的故障类型进行准确分类,提出一种结合随机森林算法分类器的故障分类方法。针对谐振接地系统下故障电流的特点,提出通过集成经验模态分解(EEMD)提取故障线路电流的暂态高频分量,建立用于区分不同故障类型的7个故障特征量。在PSCAD/EMTDC中建立仿真模型,通过改变过渡电阻、故障初始角、故障线路获得大量样本,对随机森林分类器进行训练和测试。测试结果表明所提方法具有高分类准确率。In order to accurately classify the fault types after short-circuit faults on distribution lines,this paper proposed a fault classification method combined with a random forest algorithm classifier.According to the characteristics of the fault current in the resonant grounding system,an integrated empirical mode decomposition(EEMD)is used to extract the transient high-frequency components of the fault line current,and seven fault characteristic quantities are established to distinguish different fault types.A simulation model was established in PSCAD/EMTDC,and a large number of samples were obtained by changing the transition resistance,the initial angle of the fault,and the fault line.The random forest classifier was successively trained and tested.The test results show that the method has high classification accuracy.

关 键 词：配电线路 短路故障 故障分类 高频分量 集成经验模态分解 

分 类 号：TM726[电气工程—电力系统及自动化]