基于数据挖掘的广域继电保护故障定位技术优化  

Optimization of wide-area relay protection fault location technology based on data mining

作　　者：赵凌 赵武智 齐雪雯 熊学海 万春竹 郑文龙 杜硕 ZHAO Ling;ZHAO Wuzhi;QI Xuewen;XIONG Xuehai;WAN Chunzhu;ZHENG Wenlong;DU Shuo(Guizhou Power Grid Co.,Ltd.Power Dispatching Control Center,Guiyang 550002,China;Beijing Join Bright Digital Power Technology Co.,Ltd.,Beijing 100085,China)

机构地区：[1]贵州电网有限责任公司电力调度控制中心,贵州贵阳550002 [2]北京中恒博瑞数字电力科技有限公司,北京100085

出　　处：《粘接》2024年第12期178-182,共5页Adhesion

基　　金：贵州电网有限责任公司科技项目(项目编号:GZKJXM20210483)。

摘　　要：针对广域继电保护信息量过大,影响故障定位精度的问题,设计基于数据挖掘的广域继电保护故障定位方法。利用广域继电保护结构中的智能电子装置IED,采集变电站所有元件及线路运行信息;依据粗糙集理论建立采集信息的粗糙集信息系统,通过故障条件属性及决策属性对粗糙集信息系统数据进行约简,挖掘所有与故障属性相关的数据,获取广域继电保护原始故障决策表;采用遗传算法搜索优化故障决策表,依据优化后决策表中的关联属性决策,定位主保护与后备保护故障,并及时采取相应继电保护措施。实验结果表明,该方法可以精准定位广域继电保护故障,不同类型继电保护故障定位误差均低于0.5%。In order to solve the problem that the amount of information in wide-area relay protection is too large and affects the accuracy of fault location,a fault location method for wide-area relay protection based on data mining was designed.The intelligent electronic device IED in the wide area relay protection structure was used to collect the operation information of all components and lines in the substation.According to the rough set theory,the rough set information system of the collected information was established,the data of the rough set information system was reduced through the fault condition attributes and decision attributes,all the data related to the fault attributes were mined,and the original fault decision table of wide-area relay protection was obtained.The genetic algorithm was used to search and optimize the fault decision table,and the main protection and backup protection faults were located according to the related attribute decision in the optimized decision table,and the corresponding relay protection measures were taken in time.The experimental results showed that this method could accurately locate wide area relay protection faults,and the fault location errors of different types of relay protection were less than 0.5%.

关 键 词：数据挖掘 广域继电保护 故障定位 粗糙集 遗传算法 

分 类 号：TM77[电气工程—电力系统及自动化] TP277[自动化与计算机技术—检测技术与自动化装置]