电力信息与通信技术2025，Vol.23Issue(6)：58-63,6.DOI:10.16543/j.2095-641x.electric.power.ict.2025.06.08
一种基于张量分解的电网流量异常检测方法
A Method for Detecting Abnormal Flow in Power Grid Based on Tensor Decomposition
摘要
Abstract
This paper proposes a flow anomaly detection method based on tensor decomposition to address the issue of network security threat detection in the power system.This method constructs a three-dimensional tensor based on the dimensional characteristics of power system network traffic data in time,space,and period.By utilizing the low rank characteristics of data and the structural characteristics of abnormal data,a power grid flow data anomaly detection model is constructed by constraining the structural sparse tensor with L2,1 norm to obtain anomaly detection results by solving the model using the proximal forward backward splitting algorithm.Algorithm analysis and simulation results show that this method has good precision and accuracy in anomaly detection,and can accurately detect security threats by detecting abnormal flow in the power system network.关键词
新型电力系统/异常检测/网络流量/张量分解/近端前后向分裂Key words
a new type of power system/anomaly detection/network traffic/tensor decomposition/proximal forward backward splitting分类
计算机与自动化引用本文复制引用
刘少君,陈寿龙,沙倚天,陈旸羚,吴越,李雪菲..一种基于张量分解的电网流量异常检测方法[J].电力信息与通信技术,2025,23(6):58-63,6.基金项目
国网江苏省电力有限公司科技项目"基于深度学习的内生免疫持续增强技术研究"(2023110). （2023110）
