基于时间序列的电源设备通信异常检测方法  

Time Series-Based Communication Anomaly Detection Method for Power Equipment

作　　者：赵爱勤 ZHAO Aiqin(Luoyang Institute of Electro-Optical Equipment,AVIC,Luoyang 471000,China)

机构地区：[1]中国航空工业集团公司洛阳电光设备研究所,河南洛阳471000

出　　处：《通信电源技术》2023年第22期150-152,共3页Telecom Power Technology

摘　　要：针对现有方法无法达到预期检测效果的问题,提出基于时间序列的电源设备通信异常检测方法。先利用流量采集装置捕获电源设备通信中的NetFlow报文信息,构建电源设备通信时间序列图,描述电源设备通信行为的多个时间序列的相互关系,然后通过挖掘时间序列图,计算出电源设备的通信异常系数,识别并检测电源设备的通信异常行为。实验结果表明,设计方法的平均百分比误差在1%以内,召回率在95%以上,电源设备的通信异常检测效果较好。Aiming at the problem that the existing methods can not achieve the expected detection effect,a time series-based communication anomaly detection method for power equipment is proposed.First,the NetFlow message information in the communication of the power device is captured by the traffic acquisition device,and the communication time series diagram of the power device is constructed to describe the relationship between multiple time series of the communication behavior of the power device.Then,the abnormal coefficient of the communication of the power device is calculated by mining the time series diagram,and the abnormal communication behavior of the power device is identified and detected.The experimental results show that the average percentage error of the design method is less than 1%,the recall rate is more than 95%,and the communication anomaly detection effect of power equipment is good.

关 键 词：时间序列 电源设备 通信异常 NetFlow报文 

分 类 号：TN86[电子电信—信息与通信工程]