基于核主成分分析的网关电能计量装置容错异常检测  

Fault-tolerant Anomaly Detection of Gateway Power Metering Devices Based on Kernel Principal Component Analysis

作　　者：洪晓伟 傅梓劼 HONG Xiaowei;FU Zijie

机构地区：[1]国网福建省电力有限公司泉州供电公司,福建泉州362000

出　　处：《今日自动化》2025年第1期139-141,共3页Automation Today

摘　　要：虚假数据注入攻击通过有意修改或注入错误的电力数据,进而引发电能计量装置出现错误的状态估计结果,其操作误差影响电力购销、电力供应等交易结算的公平性。文章利用核主成分分析对电能计量装置中的数据进行分析和处理,可准确检测出异常情况。试验结果表明,核主成分分析算法随着节点数量增加,计算时间逐渐缩短。且数据节点的数量越多,核主成分分析算法运算速度越快。False data injection attack by intentionally modifying or injecting the wrong power data,which in turn triggers the power metering device to appear the wrong state estimation results,and its operational error affects the fairness of the power purchase and sale,power supply and other transactions and settlements.Based on this,the article uses the kernel principal component analysis to analyze and process the data in the power metering device,so as to accurately detect the anomalous situation.Experimental results show that the kernel principal component analysis algorithm with the increase in the number of nodes,the computation time gradually decreases.And the more the number of data nodes,the faster the kernel principal component analysis algorithm operation speed.

关 键 词：网关电能计量装置 异常检测 核主成分分析算法 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]