基于多源异构数据的配电网故障信息挖掘与诊断  

Distribution Network Fault Information Mining and Diagnosis Based on Multi-source Heterogeneous Data

作　　者：马洲俊 樊飞龙[3] 王勇 王立伟 倪春华 陆昱 MA Zhoujun;FAN Feilong;WANG Yong;WANG Liwei;NI Chunhua;LU Yu(School of Electrical Engineering,Hohai University,Nanjing 210098,China;State Grid Nanjing Power Supply Company,Nanjing 210019,China;Department of Electrical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China;State Grid Shanghai Power Supply Company,Shanghai 200233,China)

机构地区：[1]河海大学能源与电气学院,江苏南京210098 [2]国网江苏省电力有限公司南京供电公司,江苏南京210019 [3]上海交通大学电气工程系,上海200240 [4]国网上海市电力有限公司市南供电公司,上海200233

出　　处：《供用电》2018年第8期31-39,共9页Distribution & Utilization

基　　金：国网江苏省电力公司科技项目资助(J2017074)~~

摘　　要：配电网海量运行数据为其运行分析提供了良好的基础,如何挖掘多源异构数据诊断故障是提升配电网安全运行水平和优质服务水平的关键。文章以生产管理系统、故障抢修以及SCADA数据为对象,分析多源信息构成与属性,建立SCADA事件的预集规则,并匹配关联故障抢修事件与SCADA数据。基于数据来源、属性等不同维度,综合考虑数据预集规则,对多源异构数据进行分类组合与融合,并存储于大数据hadoop平台。基于多源异构数据的配电网故障诊断首先对事件进行判断,挖掘与故障相关的保护动作、遥测、事故汇总等,明确故障对象、内容与属性,并结合辅助信息验证诊断结果。利用该策略建立的系统部署于某城市配电网,其诊断性能及正确率满足要求,可有效提高配电网数据挖掘与分析能力,为调度运行及检修提供支撑。Massive operating data provides a good foundation for the operation analysis of distribution network.How to mine multi-source heterogeneous data to diagnose faults is the key to improve the safe operation level and quality service level of the network.Based on data of the production management system,fault repair system and SCADA system,this paper analyzes the composition and attributes of multi-source information,establishes the SCADA event preset rules,and matches the associated fault repair events with the SCADA data.Based on different dimensions such as data source,and attributes,the data preset rules are comprehensively considered,and multi-source heterogeneous data is combined,merged and stored in the big data hadoop platform.The distribution network fault diagnosis based on multi-source heterogeneous data firstly judges the events and mines the protection action,telemetry,and accident level which are related to fault.The diagnosis results are verified by identifying the fault object,content,attribute and the auxiliary information.The proposed strategy is used to establish a system deployed in a certain city distribution network.Its diagnostic performance and accuracy rate meet the requirements.The proposed strategy effectively improves the data mining and analysis capabilities of distribution network,which provides support for dispatching operation and maintenance.

关 键 词：故障诊断 多源异构数据 事件预集 信息挖掘 配电网 

分 类 号：TM93[电气工程—电力电子与电力传动]