基于虚假数据注入攻击的网络安全检测

Grid Network Security Detection Based on FDIA

为提高智能电网的安全性,结合传感器量测数据和攻击向量服从正态分布的特性,提出了一种基于高斯混合模型的虚假数据注入攻击(False Data Injection Attacks,FDIA)检测方法。在该方法中,通过EM算法求解出高斯混合模型参数,然后根据判断准则,利用测试数据对高斯混合模型的分类效果进行验证。仿真实验结果表明,在IEEE-18和IEEE-30系统节点网络攻击检测中,基于高斯混合模型的FDIA检测相较于SVM的FDIA检测精度更好,但攻击强度和协方差矩阵是关键影响因素。

In order to improve the security of smart grid,combined with the characteristics that the sensor measurement data and attack vector obey normal distribution,an FDIA detection method based on Gaussian mixture model is proposed.In this method,the parameters of Gaussian mixture model are solved by EM algorithm,and then according to judgment criteria,the classification effect of Gaussian mixture model is verified by test data.Simulation results show that in the node network attack detection of IEEE-18 and IEEE-30 system,FDIA detection based on Gaussian mixture model has better detection accuracy than that based on SVM,but attack intensity and covariance matrix are the key influence factors.