大数据环境下输变电设备监控信息的统计及异常检测研究

Research on Statistics and Anomaly Detection of Monitoring Information for Power Transmission and Transformation Equipment Under Big Data Environment

针对输变电设备异常状态数据的检测问题,基于孤立森林算法提出了一种监控信号参数统计分析方法,设计了集中监控辅助决策系统。首先,分析了现有数据异常检测方法,提出了现有输变电设备状态数据检测研究存在的不足。其次,给出了监测信息数据采集及统计的总体架构,分析了智能电网的控制技术支撑系统(D5000)、调度管理系统(OMS)以及在线监控系统等关键环节的数据统计注意事项。再次,给出了孤立森林算法的实现原理,同时利用LOF、K-means算法构建了3种异常数据检测模型,通过仿真发现基于孤立森林算法建立的异常数据检测模型具有更高的精度与运算速率。最后,以500 kV输电线路为例进行了案例分析。结果表明,提出的集中集控辅助决策系统能实现对设备异常状态的准确识别。

A monitoring signal parameter statistical analysis method based on isolated forest algorithm is proposed for the detection of abnormal status data of power transmission and transformation equipment,and a centralized monitoring auxiliary decision-making system is designed.Firstly,an analysis was conducted on existing data anomaly detection methods,and the shortcomings in existing research on status data detection of power transmission and transformation equipment were pointed out.Secondly,the overall architecture of monitoring information data collection and statistics was provided,and the data statistics precautions for key links such as the control technology support system(D5000),dispatch management system(OMS),and online monitoring system of the smart grid were analyzed.Then,the implementation principle of the isolated forest algorithm was presented,and three anomaly data detection models were constructed using LOF and K-means algorithms.Through simulation,it was found that the anomaly data detection model based on the isolated forest algorithm has higher accuracy and computational speed.Finally,a case study was conducted on a 500 kV transmission line.The results showed that the proposed centralized control assisted decision-making system can accurately identify equipment abnormal states.