基于历史数据挖掘的DSCADA告警数据治理方法  

Alarm data governance method for DSCADA based on historical data mining

作　　者：郭峰 姜宽 郭秉涛 杨振睿 蔡斌 周鑫 GUO Feng;JIANG Kuan;GUO Bingtao;YANG Zhenrui;CAI Bin;ZHOU Xin(State Gird Shanghai Municipal Electric Power Company,Shanghai 200080,China)

机构地区：[1]国网上海市电力公司市区供电公司,上海200080

出　　处：《供用电》2024年第5期72-79,共8页Distribution & Utilization

基　　金：国网上海市电力公司科技项目(B30911230005)。

摘　　要：针对电网企业配电网数据采集与监视控制(distribution supervisory control and data acquisition,DSCADA)系统存在的告警信号频发问题,提出了基于历史数据挖掘的DSCADA告警数据治理方法。首先,利用基于字符串匹配的分词算法对DSCADA存储的10 kV配电站原始告警信号进行文本分词,实现对DSCADA主站系统存储的海量历史告警记录的多维统计;然后,通过构建10 kV配电站告警知识图谱,辅助DSCADA系统异常告警信号的定位、分析过程;最后,提出了基于历史告警记录分析的DSCADA告警配电站健康度评价方法,用于指导运维人员消缺。所提方法改变了电网企业原有仅靠人工管控的技术现状,为DSCADA运行数据治理提供了一种高效的技术分析手段。Aiming at the issue of frequent occurrence of alarm signals of distribution supervisory control and data acquisition(DSCADA),this paper proposes an alarm data governance method for DSCADA based on historical data mining.Firstly,the string-matching based text segmentation algorithm is applied to the segmentation of 10 kV distribution substations’original alarm signals stored in DSCADA,the massive alarm records stored in DSCADA are analyzed multi-directionally.Meanwhile,the knowledge graph between 10 kV distribution substations and alarm signals is established,which helps assist in the location and analysis of abnormal alarm signals in DSCADA.Finally,the evaluation method of health indexes for distribution substations is proposed based on the analysis of historical alarm signals of DSCADA.The evaluation results can be applied to the defeat elimination of distribution substations.The proposed method can change the previous artificial defeat elimination situation for electric power companies,which provides an efficient technical approach for the data governance of DSCADA.

关 键 词：配电自动化 配电网数据采集与监视控制 告警频发 数据挖掘 数据治理 

分 类 号：TM76[电气工程—电力系统及自动化]