基于数据挖掘的配网数字化计量系统运行特性监测和评价方法研究

On⁃site Monitoring and Evaluation Method of Operation Characteristics of MV Distribution Digital Metering System Based on Data Mining Technique

准确评估配网数字化计量系统在电力系统运行过程中的计量性能,提升其运行中的准确度、稳定性和可靠性是其未来用于电能贸易结算的关键问题。目前系统中投运的配网数字化计量系统存在故障率高、故障原因难以评估等困难。文中提出一种基于在线监测和数据挖掘的配网数字化计量系统运行状态评价方案,通过在线监测装置采集和存储运行数据,再通过拉伊达准则对原始数据进行预处理,利用决策树算法对高维数据进行降维和排序,能快速准确评估误差主要来源。利用文中提出的运行特性评价方法,对配网数字化计量试点运行的两类数字化计量系统的运行数据进行监测和分析,验证文中提出的方法的有效性,通过分析一年的有功电量误差获得两种数字化计量系统设计方案的优缺点,为数字化计量系统未来的应用提供了实践和选型的依据。

In order to improve the stability and reliability of the electronic instrument transformers(EITs)in energy trade,it is necessary to accurately assess the calculating characteristics of the EITs in operation of electric power sys⁃tem.However,the MV digital metering system in the electric power system has some problems,such as high failure rate and difficult fault location.In this paper,a new scheme based on on⁃site monitoring and data mining technique is proposed,in which the Rajda criteria and decision tree algorithm are used to delete the outline value and reduce the feature dimension in order to evaluate main sources of error.Using this new data mining technique,the operation data of two types of digital metering systems in the pilot distribution network are monitored and analyzed,and the results verify the effectiveness of the scheme.The advantages and disadvantages of the two digital metering systems are summed up through analyzing the error of active power in one year.This research may provide reference for future ap⁃plication of digital metering system.