基于移动小波树的电力监控异常数据自动识别算法研究  

Research on automatic recognition algorithm of abnormal data in power monitoring based on mobile wavelet tree

作　　者：夏景 梁薇 吴珠瑛 XIA Jing;LIANG Wei;WU Zhu-ying(State Grid Shanghai Songjiang Electric Power Supply Company,Shanghai 201699,China)

机构地区：[1]国网上海市电力公司松江供电公司,上海201699

出　　处：《电子设计工程》2020年第18期148-152,共5页Electronic Design Engineering

基　　金：国网上海松江供电公司2019年实体营业厅转型升级项目(B2093519000Q)。

摘　　要：为保证电力系统正常运行,以防出现窃电、漏电等异常用电情况,进行电力监控异常数据自动识别具有重要的现实意义。针对基于数据挖掘、机器学习以及统计的3种数据异常识别算法无法应对不同长度的数据,导致识别准确性、效率较低的问题,提出一种基于移动小波树的电力监控异常数据自动识别算法。该算法分为两部分内容:先是对采集到的用电数据进行处理,包括数据脱敏、数据填补、颠簸去除等步骤,提高数据质量,后构建移动小波树结构,实现异常数据自动识别。结果表明:与基于数据挖掘、机器学习以及统计的三种数据异常识别算法相比,所提算法的准确性和效率均较高,解决了传统算法存在的问题。In order to ensure the normal operation of the power system,in order to prevent the occurrence of electricity theft,leakage and other abnormal power use,it is of great practical significance to automatically identify the abnormal data of power monitoring.In view of the problem that the three anomaly recognition algorithms based on data mining,machine learning and statistics can't deal with different length of data,resulting in low accuracy and efficiency of recognition,an automatic anomaly recognition algorithm for power monitoring based on mobile wavelet tree is proposed.The algorithm is divided into two parts:first,the collected power consumption data is processed,including data desensitization,data filling,bump removal and other steps to improve the data quality,and then the mobile wavelet tree structure is constructed to realize the automatic recognition of abnormal data.The results show that compared with the three algorithms based on data mining,machine learning and statistics,the proposed algorithm has higher accuracy and efficiency,and solves the problems of the traditional algorithm.

关 键 词：移动小波树 电力监控 异常数据 识别算法 

分 类 号：TN274[电子电信—物理电子学]