Online Detection of False Data Injection Attacks to Synchrophasor Measurements: A Data-Driven Approach
Online Detection of False Data Injection Attacks to Synchrophasor Measurements: A Data-Driven Approach
复制标题
同步相量测量的虚假数据注入攻击的在线检测：数据驱动的方法
DOI：
--
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
Le Xie
中科院分区：
文献类型：
--
作者：
Meng Wu;Le Xie
本文提出了一种在线数据驱动算法来检测针对同步相量测量的虚假数据注入攻击。该算法应用基于密度的局部离群因子（LOF）分析来检测数据中的异常，这些异常可以被描述为来自网格的所有同步相量测量值之间的时空离群值。通过利用同步相量测量的多个时刻之间的时空相关性，该方法可以检测使用从单个快照获得的测量无法检测到的虚假数据注入攻击。该算法不需要系统参数或拓扑结构的先验知识。计算速度显示出令人满意的潜力，在线监测应用。仿真和实际同步相量数据的算例验证了算法的有效性。
This paper presents an online data-driven algorithm to detect false data injection attacks towards synchronphasor measurements. The proposed algorithm applies density-based local outlier factor (LOF) analysis to detect the anomalies among the data, which can be described as spatiotemporal outliers among all the synchrophasor measurements from the grid. By leveraging the spatio-temporal correlations among multiple time instants of synchrophasor measurements, this approach could detect false data injection attacks which are otherwise not detectable using measurements obtained from single snapshot. This algorithm requires no prior knowledge on system parameters or topology. The computational speed shows satisfactory potential for online monitoring applications. Case studies on both synthetic and realworld synchrophasor data verify the effectiveness of the proposed algorithm.