【摘要】 智能电网容易受到虚假数据注入攻击。对虚假数据的检测有助于提高电网状态估计准确性,从而提高智能电网的稳定性。为了能自动检测出智能电网中的虚假注入数据,提出一种基于双低秩约束的智能电网虚假数据检测算法。与已有算法相比,该算法不使用训练数据自身作为低秩学习字典,而是引入一个低秩特征转换矩阵,使得用于低秩学习的字典具有更好的灵活性,以及学习到的模型具有更强的泛化能力。该算法可应用于智能电网的电力状态估计。在多个IEEE节点测试系统上的实验表明该算法比目前广泛使用的主成分分析算法和鲁棒主成分分析算法具有更好的性能。更多还原

【Abstract】 Smart grid is vulnerable to False Data Injection Attacks( FDIA). The detection of false data helps to improve the accuracy of power state estimation and the stability of smart grid. To automatically detect the false data from smart grid,this paper proposed a new algorithm based on dual low-rank constraint. Compared with existing approaches, the proposed method does not deploy the training data itself as the learning dictionary. Instead, it introduced a feature transformation matrix to enhance the flexibility of the dictionary and the generalization ability of the learned model. The proposed method can be used for power state estimation in smart grid. At last, experiments on several IEEE bus systems verify that the proposed method is much better than existing ones, e. g., PCA( Principal Component Analysis), Robust PCA and LRDE( Low-Rank Discriminant Embedding).更多还原