在DBSCAN+LOF的大扰动工况下PMU装置不良数据检测算法研究  

Research on Bad Data Detection Algorithm of PMU Device under Large Disturbance Condition of DBSCAN+LOF

作　　者：陈涛 张水喜 袁正华 黄敏 王建军 CHEN Tao;ZHANG Shuixi;YUAN Zhenghua;HUANG Min;WANG Jianjun(State Grid Henan Electric Power Company,Zhengzhou 450000,China;Luohe Power Supply Company of State Grid Henan Electric Power Company,Luohe 462000,China)

机构地区：[1]国网河南省电力公司,河南郑州450000 [2]国网河南省电力公司漯河供电公司,河南漯河462000

出　　处：《微型电脑应用》2024年第6期74-78,共5页Microcomputer Applications

基　　金：基于PMU数据的电力系统安全校核技术研究与应用(5217G020000M)。

摘　　要：针对传统k-means算法异常点检测算法在大扰动情况下易产生误检、误判的问题,提出基于DBSCAN+LOF的电力系统PMU不良数据检测算法。结果表明:PMU正常数据存在较强的时空相似性,PMU不良数据的时空相似性均较弱,大扰动PMU数据存在较强的空间相似性,但时间相似性较弱;根据3种数据的时空特征,可利用DBSCAN算法检测出异常点,再利用LOF算法计算局部离群因子,通过局部离群因子大小来判别大扰动PMU数据和PMU不良数据;将提出的算法应用到电力系统短路故障中,结果显示在短路故障发生和切除时刻,LOF计算结果显示为大扰动PMU数据,在故障切除后,LOF计算结果显示为PMU不良数据,检测结果与实际情况完全相符,算法是合理有效的。In view of the problem that the traditional k-means algorithm for outlier detection is prone to false detection and false judgment in the case of large disturbances,a DBSCAN+LOF based algorithm for detecting bad data of PMU in power system is proposed.The results show that the normal PMU data have strong spatial and temporal similarity,the poor PMU data have weak spatial and temporal similarity,the large disturbance PMU data have strong spatial similarity,but weak temporal similarity.According to the space-time characteristics of the three types of data,the DBSCAN algorithm is used to detect outliers,the LOF algorithm is used to calculate local outlier factors,and the large disturbance PMU data and PMU bad data can be discriminated by the size of local outlier factors.The algorithm proposed is applied to the short-circuit fault of the power system.The results show that at the time of occurrence and removal of the short-circuit fault,the LOF calculation results are shown as large disturbance PMU data.After the fault is removed,the LOF calculation results are shown as PMU bad data.The detection results are completely consistent with the actual situation.The algorithm is reasonable and effective.

关 键 词：电力系统 PMU不良数据 大扰动 检测算法 DBSCAN LOF 

分 类 号：TM764.1[电气工程—电力系统及自动化]