电力设备多参量监测数据清洗研究现状及展望

Research Review and Prospect of Data Cleaning for Multi-parameter Monitoring Data of Power Equipment

基于电力设备多参量监测数据,开展电力设备态势感知工作是提高设备检修效率、消除故障隐患、保障电力系统安全稳定运行的重要途径之一。然而,数据采集、传输与存储过程中受到的各类干扰导致原始监测数据中存在大量的偏差与缺失,进而影响态势感知准确性,亟须通过数据清洗提升数据质量。在综合分析电力设备多参量监测数据清洗领域文献的基础上,概述了电力设备多参量监测数据质量影响因素。对电力设备多参量监测数据清洗的通用框架进行了总结,该框架包括多参量相关性分析、异常数据检测、异常数据分类和“脏”数据修复4个环节,对各环节的常用方法进行了对比分析,并介绍了特殊应用场景下的数据清洗方式。探讨了提升数据清洗效率的2种方式,分析了电力设备多参量监测数据清洗研究领域面临的主要挑战,对未来发展趋势进行了展望。

The situation awareness of power equipment based on multi-parameter monitoring data is one of the important approaches to enhance equipment maintenance efficiency,eliminate hidden faults,and ensure the safe and stable opera-tion of the power system.However,various interferences during the data collection,transmission,and storage processes lead to significant deviations and missing data in the original monitoring data,which in turn affects the accuracy of situa-tional awareness.Therefore,it is urgent to perform data cleansing to enhance data quality.On the basis of comprehensively analyzing the literature in the field of multi-parameter monitoring data cleaning of power equipment,we analyzed the influencing factors of data quality for power equipment multi-parameter monitoring,and we summarized the general framework for data cleaning of multi-parameter monitoring data,including multi-parameter correlation analysis,abnormal data detection,abnormal data classification,and“dirty”data repair.Moreover,the common methods of each part are compared and analyzed,and the data cleaning methods under special scenarios are introduced.Additionally,two approaches to improve the efficiency of data cleaning are explored.Finally,we identify the main challenges of data cleansing field for multi-parameter monitoring data of power equipment and provide an outlook on future development trends.